Binge Drinking During Pregnancy

Transcription

1 Katrine_PhD_forbag 21/02/09 18:17 Side 1 Binge Drinking During Pregnancy Methodological issues and short-term consequences of intrauterine exposure PhD thesis Faculty of Health Sciences University of Copenhagen KATRINE STRANDBERG-LARSEN National Institute of Public Health University of Southern Denmark 2008

2 Katrine_PhD_forbag 21/02/09 18:17 Side 2

3 BINGE DRINKING DURING PREGNANCY Methodological issues and short-term consequences of intrauterine exposure PhD thesis Faculty of Health Sciences University of Copenhagen, Denmark By Katrine Strandberg-Larsen, M.Sc. in Public Health National Institute of Public Health University of Southern Denmark, Denmark Academic advisors Morten Grønbæk, Professor, PhD, Dr Med Sci National Institute of Public Health University of Southern Denmark, Denmark Anne-Marie Nybo Andersen, Professor, PhD Division of Epidemiology University of Southern Denmark, Denmark Per Kragh Andersen, Professor, PhD, Dr Med Sci Department of Biostatistics University of Copenhagen, Denmark

4 T H I S T H E S I S I S B A S E D O N S I X P A P E R S : Paper I Strandberg-Larsen K, Nybo Andersen AM, Olsen J, Nielsen NR, and Grønbæk M: Do women give the same information on binge drinking during pregnancy when asked repeatedly? European Journal of Clinical Nutrition 2006; 60: Paper II Strandberg-Larsen K, Nielsen NR, Nybo Andersen AM, Olsen J, and Grønbæk M: Characteristics of women who binge drink before and after they become aware of their pregnancy. European Journal of Epidemiology 2008; 23: Paper III Strandberg-Larsen K, Nielsen NR, Grønbæk M, Andersen PK, Olsen J, and Nybo Andersen AM: Binge drinking in pregnancy and risk of fetal death. Obstetrics & Gynecology 2008; 111: Paper IV Strandberg-Larsen K, Grønbæk M, Nybo Andersen AM, Andersen PK, and Olsen J: Alcohol drinking pattern during pregnancy and risk of infant mortality. Accepted in Epidemiology in a revised version. Paper V Strandberg-Larsen K, Skov-Ettrup LS, Grønbæk M, Nybo Andersen AM, Olsen J, and Tolstrup J: Alcohol drinking pattern during pregnancy and congenital heart disease. Submitted. Paper VI Strandberg-Larsen K, Jensen MS, Ramlau-Hansen CH, Grønbæk M, and Olsen J: Alcohol binge drinking during pregnancy and cryptorchidism. Submitted.

5 P R E F A C E The work presented in this PhD thesis was carried out between 2005 and 2008 at the National Institute of Public Health, University of Southern Denmark. It was supported by grants from the Danish Graduate School of Public Health, The Health Insurance Foundation and the Ministry of the Interior and Health. I wish to express my sincere gratitude to my academic advisors Morten Grønbæk, Anne-Marie Nybo Andersen and Per Kragh Andersen for sharing their scientific insight and experience with me. Especially, I owe thanks to Morten and Anne- Marie for their encouragement and never-ending confidence in me when needed, and to Per for always being willing to help me on statistical issues and for his sharp eyes on my description of my use of statistical methods. I have also had the great pleasure to collaborate with Jørn Olsen who has contributed with many constructive ideas and criticisms, and not least let me share his office at UCLA for which I am truly grateful. The Augustinus Foundation, the Knud Højgaard Foundation and the Faculty of Health Sciences, University of Copenhagen generously supported my research stay at UCLA. I would also like to thank all my other co-authors for their valuable contributions to the papers that are included in this thesis and all of my great colleagues at the National Institute of Public Health for contributing to make the institute a pleasant and stimulating place to work. I am grateful to my good friends and colleagues Naja Hulvej Rod and Mette Juhl for taking the time to comment on my thesis and for always being willing to discuss epidemiological issues with me. My gratitude also goes to every single participant in the Danish National Birth Cohort (DNBC), to all the people who have been involved in establishing and coordinating the cohort, as well as to those who contributed to the construction of the DNBC research database. Last, but not least my most sincere and deepest gratitude goes to my family and friends for their love and continuous support and a special thanks to Martin and Emilie for daily inspiration. Katrine Strandberg-Larsen Copenhagen, December 2008

6

7 T A B L E O F C O N T E N T S 1. I N T R O D U C T I O N Evidence from animal models Evidence from studies in humans A I M S M A T E R I A L A N D M E T H O D S The Danish National Birth Cohort Measures of binge drinking and average alcohol intake Handling incomplete information on timing of binge drinking Assessment of endpoints Use of causal diagrams to select confounders Statistical methods R E S U L T S Repeatability of information on binge drinking (Paper I) Characteristics of pregnant women who reported binge drinking (Paper II) Binge drinking in pregnancy and fetal death (Paper III) Alcohol drinking pattern and infant death (Paper IV) Alcohol drinking pattern and congenital heart disease (Paper V) Alcohol drinking pattern and cryptorchidism (Paper VI) Population attributable fractions D I S C U S S I O N Summary of main findings Consistency with previous studies Strengths and weaknesses C O N C L U S I O N A N D P E R S P E C T I V E S Conclusion Public health implications S U M M A R Y I N E N G L I S H S U M M A R Y I N D A N I S H R E F E R E N C E S P A P E R S

8

9 1. I N T R O D U C T I O N Alcohol is a human teratogen and the link between heavy drinking during pregnancy and the risk of fetal alcohol syndrome is well established. 1-3 Intake of one or more alcoholic drinks per day during pregnancy has consistently been shown to be associated with spontaneous abortion, 4-8 reduced birth weight, 9;10 preterm delivery 11;12 as well as the development of the child However, it remains unsettled whether smaller amounts of alcohol during pregnancy may harm the developing fetus Most studies have assessed the risk of drinking alcohol in pregnancy on the basis of measures of average intake, but animal models suggest, however, that the effects of alcohol depend upon the peak blood alcohol concentrations rather than on the average intake Thus, when assessing damages caused by prenatal alcohol exposure it may be important to assess the effects of high blood alcohol concentrations, which are achieved by intake of large volumes of alcohol on a single occasion, often referred to as binge drinking. Binge drinking is most often defined as intake of at least five alcoholic drinks on a single occasion. 30;31 The effect of binge drinking in humans independently of average alcohol intake has not been widely reported, 32 but binge drinking has been highlighted as a potentially teratogenic way to consume alcohol. 33 Prenatal alcohol exposure targets distinct structures when administered at different developmental timepoints Furthermore, the duration of the binge episodes is most often limited to a few hours and the short half-life of ethanol makes the fetal exposure very limited in time. Thus, the effects of binge drinking during pregnancy are likely to be time-specific and depend upon the stage of the embryonic development at the time of binge drinking. 38;39 It may therefore be important not only to include information about intermittently high levels of alcohol and the number of such episodes, but also on when in pregnancy these episodes took place. 31;38;40 Government recommendations regarding alcohol use for pregnant women or for women who are planning pregnancy differ between countries; some, such as Australia and New Zealand, recommend moderation, whereas others, such as the USA, France, Norway, and Sweden, recommend abstinence. The current UK Department of Health guidelines recommend that women, who are trying to become pregnant or are at any stage of pregnancy, should avoid drinking alcohol, but if the women choose to drink, they should not drink more than one or two units of alcohol once or twice a week, and they should avoid episodes of intoxication. 41 These guidelines were in accordance with the recommendations from 1999 from the Danish National Board of Health that recommended that pregnant women should, if possible, avoid intake of alcohol, and if deciding to drink, they should never drink more than 1 drink per 24 hours, and they should not drink every day. 42 In 2007, these recommendations were changed back to the advice given before 1999: to stay completely abstinent during pregnancy. 43 It is worth mentioning, that the drink size in the two countries differs; in the UK, one unit of alcohol contains around 8 grams of alcohol, whereas a drink in Denmark contains around 12 grams of alcohol. 44 Both the current UK guidelines and the Danish recommendations from 1999 explicitly state that women should avoid episodes of binge drinking, and avoidance of binge drinking is also implicitly included in the recommendations of complete abstinence. 1

10 Binge drinking is an important topic for women s health, as well as for the fetus and child. The majority of women do not change their drinking habits before the time when pregnancy is recognized, and binge drinking may therefore occur in the pre-recognized part of pregnancy. In Norway and Denmark around 25 and 50%, respectively, of women have binged at least once while being pregnant. 31;48 This may be problematic for several reasons. First, binge drinking is a drinking pattern that is on the increase also among women of childbearing age Second, the potential of adverse effects on the fetus, and, third, the fact that the majority (85%) of pregnant women consider binge drinking during pregnancy to be harmful. 52 Binge drinking in the unrecognized part of pregnancy may, therefore, be a cause for concern, and health professionals are often consulted about the possible detrimental effects of binge drinking during early pregnancy. Clinical advice should, whenever possible, be based on evidence. 53 Therefore, what is the evidence for the health effects of binge drinking during pregnancy? In the next two sections the evidence from animal studies and epidemiologic studies will be described Evidence from animal models Alcohol rapidly diffuses the placenta and the blood alcohol concentration of the fetus reaches the same level as the maternal blood alcohol concentration within an hour subsequent to intake It is unclear whether the damaging effects of prenatal exposure to alcohol are due mainly to ethanol per se, or its primary metabolite acetaldehyde It appears that prenatal alcohol exposure may affect all stages of fetal development: chromosome damage, cell death, cell division and proliferation, cell growth and differentiation, cell migration and adhesion. 57;59-61 The mechanism of action for the detrimental effects of alcohol on these cellular events has not been fully clarified, but many possible mechanisms of actions have been suggested, e.g. increased oxidative stress, insufficient transfer of nutrients across the placenta, alteration of the production of hormones and diminished oxygen delivery to fetal tissues. 57;61;62 Animal models have suggested that it is the peak blood alcohol concentration rather than the total volume consumed that determines the level of damage, and that the neural system is particularly susceptible to high blood alcohol concentrations. 25;27;63-67 The concentration of alcohol in the blood is not solely a product of the dose of alcohol and the rate of consumption, 65 but also depends on e.g. body size, gender, concurrent use of other drugs, age, genetic factors, and the amount of time elapsed between eating and drinking However, the peak blood alcohol concentrations achieved with binge-like drinking are higher than when an equal volume is consumed more slowly. 63;70 An additional consequence of binge drinking is prolonged alcohol exposure, because the rate of alcohol metabolism remains the same regardless of the amount consumed. As a result, the higher the blood alcohol concentration, the longer it takes for the body to metabolize all the alcohol. 65;67;71 Research on rodents have suggested that a single binge episode or period of binge exposure is sufficient to produce congenital malformations, including craniofacial malformations that resemble those seen in humans with the fetal alcohol syndrome, increase in resorptions/fetal deaths, growth retardation, skeletal dysmorphogenesis, microcephaly, reduced brain weight and loss in regional populations of neurons ;29;30;63;65;72-78 Further, the regional loss of neurons seems to depend upon the 2

11 time of alcohol exposure. 78;79 Nonhuman primates have also been used to study the relation between weekend binge drinking (~once a week) and potential health consequences. 27;28;71;80-86 The findings from these studies suggest that binge drinking during early pregnancy increases the risk of spontaneous abortion. 71;80;83 Furthermore, binge-like alcohol consumption once a week during pregnancy is capable of producing facial anomalies, growth deficiency or central nervous system dysfunction, and cognitive abnormalities may be observed in the absence of physical anomalies. 27;28;81;82;84 Results from this nonhuman primate model have also suggested that binge drinking during early pregnancy followed by later gestational abstinence may produce a similar amount of gross brain damage and cognitive deficits as binge drinking throughout pregnancy. 27;86 Furthermore, the results suggest that there may be a critical period for induction of alcohol-induced craniofacial alterations that occurs very early in pregnancy and is of very short duration. 86 More recent research using an ovine model likewise indicates that binge drinking may produce loss of neurons and alter the bone strength and growth Taken together, the findings from animal models clearly demonstrate that binge drinking may be more harmful than if the same or even higher volumes of alcohol are consumed more slowly, and that the effect may be time-specific. 30;67 However, it remains uncertain whether the findings from animals can be extrapolated to humans Evidence from studies in humans Few epidemiological studies have addressed the potential adverse effects of prenatal exposure to binge drinking independently of average alcohol intake. The most consistent finding from the existing epidemiological literature is a possible relation to detrimental effects on the brain development. 32 Prenatal exposure to binge drinking has been shown to be associated with disinhibited behavior, reduction in verbal IQ, increase in delinquent behavior, learning disabilities, deficits in attention, memory, reading, and cognitive processing, visible brain alterations on electroencephalogram, and psychiatric disorders. 17; However, two studies have reported no association between binge exposure and child development during the first 3-4 years of age. 93;104 Preliminary analyses on a sub-sample of the Danish National Birth Cohort also suggest no association between prenatal exposure to binge drinking and information processing time and attention at age five. 105;106 The findings on birth weight are more inconsistent as around half of the studies indicate an association, while the other half reported no association. 9;94; The studies that indicated a positive association were either not fully adjusted for possible confounders 94;107;108 or it was impossible to separate out the effect of binge drinking from heavy drinking. 9 In relation to other outcomes most studies have found sparse, if any, association with spontaneous abortion, 4;114 length at birth, 109;114 head circumference, 109; gestational age, 109 apgar score, 109 hospitalization with asthma in childhood, 115 craniofacial features related to the fetal alcohol syndrome, 114;116 and malformations. 107; However, other studies on malformations have indicated an association with the mean number of abnormalities at birth 123 and absent kidney development. 124 A single study has shown a positive association between prenatal exposure to binge drinking and sudden infant death

12 Altogether, at the present time, no consistent conclusions can be made about whether binge drinking affects fetal health. The most consistent finding, from the animal models and the epidemiological literature, is a probable detrimental effect on neurodevelopmental outcomes. However, it is important to emphasize that relatively few human studies have examined the association between binge drinking and adverse pregnancy and child outcomes. Furthermore, the existing studies were most often either of limited size, did not include information on the number and timing of binge episodes, or were based on women with a substantial average intake. 31;32 Thus, research with information on the number and timing of binge drinking independently of a high average intake is needed as a basis for evidence-based guidelines or advice from clinicians to pregnant women who fear for the health of their fetus because they have binged early in pregnancy. 32 4

13 2. A I M S The overall aim of the thesis was to address a potential relation between prenatal exposure to binge drinking and short-term health consequences for the offspring. Four fetal health outcomes were selected to reflect some of the possible consequences of intrauterine exposure to binge drinking. These potential relations were addressed in studies based on data from the Danish National Birth Cohort, a nation-wide cohort of pregnant women and their offspring. Initially, the repeatability of the obtained information on the number and times of binge drinking was addressed and the women who binged during pregnancy were characterized. Six studies form the basis of this thesis and the specific aims in these studies were: Paper I Paper II To examine if women report the same number and times of binge drinking when asked repeatedly To identify lifestyle, socio-demographic factors and aspects of reproductive history that are associated with binge drinking in the preand post-recognized part of pregnancy Paper III To examine the association between binge drinking during the first 16 weeks of gestation and the risk of early as well as late fetal death Paper IV Paper V Paper VI To examine the association of maternal average alcohol consumption and binge drinking with risk of infant mortality divided into the neonatal and post-neonatal period To examine the association between alcohol drinking pattern during pregnancy and congenital heart disease, specifying the isolated forms: ventricular and atrial septal defects To examine if prenatal exposure to binge drinking, especially exposure during critical stages of the testis development, is associated with the risk of cryptorchidism 5

14 3. M A T E R I A L A N D M E T H O D S 3.1. The Danish National Birth Cohort The Danish National Birth Cohort (DNBC) is a nation-wide cohort of about 100,000 pregnant women and their offspring. 126 It was established to facilitate large-scale epidemiologic studies of short-term and long-term consequences of intrauterine exposures. Enrollment into the cohort took place between March 1996 and November General practitioners introduced the study to women at their first contact to the antenatal care system, which is usually scheduled at gestational weeks 6 to 12. The women were included as participants when the study centre received a signed informed consent form. About 50% of all general practitioners in Denmark participated in the enrollment into the cohort and about 60% of the invited women returned the informed consent form. To be eligible to participate in the DNBC the women had to be pregnant, not planning an induced abortion, have a permanent address in Denmark, and speak Danish well enough to be interviewed by telephone in Danish. Information on exposures during pregnancy, the postnatal period and the development of the children was collected in four computer-assisted telephone interviews taking place in approximately pregnancy week 17 (quartile: 14-20), week 32 (quartile: 30-34), and at six months (quartile: 5-6) and 18 months (quartile: 18-19) after the delivery. Some of the fetuses did not survive until the time of the first interview, and if a woman when contacted for this interview stated that she was no longer pregnant she was offered to take part in an early case interview. The questions in the early case interview were equivalent to those in the first interview, but some of the wording was changed and a few questions, like general statements about psychological well being in pregnancy were omitted. Instead the early case interview included questions to describe the circumstances of the fetal loss and concerned only exposures during the first 16 weeks of pregnancy. The ethics committee for Copenhagen and Frederiksberg approved the DNBC ((KF) /94) and the Danish Data Protection Board approved the use of data for the work included in this thesis ( ). The steering committee for the DNBC also approved our use of data to study the health consequences of binge drinking during pregnancy ( ). Data was restricted in different ways in the different studies. To study the repeatability of information on binge drinking it was essential to have information from at least two of the interviews. The data cleaning process was ongoing when we made the analyses for Paper I, and data from the first interview was only available on approximately 86,500 participants, instead of the finale 90,165. In paper I we restricted the data to two subgroups. The first consisted of 76,307 participants with information on binge drinking from the first and second interviews, and the second subgroup consisted of 8,933 women, for whom we had information on binge drinking during pregnancy weeks 30 to 36 from the second and third interviews. Figure 1 shows how we restricted data to Papers II and III. In brief, we included women with information on exposures during the early part of pregnancy and informative data on binge drinking. The rest of the papers were based on data that was restricted to women who gave birth to a liveborn singleton and completed the first interview and provided informative data on binge drinking and gestational age, see Figure 2. 6

15 Figure 1. The restriction of data for the studies in Papers II and III STUDY POPULATION FIRST INTERVIEW DATA EXCLUSIONS OF PARTICIPANTS ELIGIBLE FOR ANALYSIS 90,165 1 st pregnancy interview Jun 27, 1997 Feb 7, Missing on binge drinking 836 Missing on timing of binge drinking 3,652 Missing on other variables in the model 85,334 Paper II 50 Hydatidiform mole/electopic pregnancy 2,552 Early case interview Jun 2, 1997 Mar 29, Missing on gestational age 367 Missing on binge drinking 3,059 Missing on other variables in the model 89,201 Paper III 100,418 Enrolled in DNBC Mar 18, 1996 Nov 1, ,636 Interview data missing 65 No interview in design due to hydatidiform mole or induced abortion related to maternal illness

16 Figure 2. The restriction of data for the studies in Papers IV to VI STUDY POPULATION PREGNANCY OUTCOME INTERVIEW DATA PARTICIPANTS FULFILLING THE INCLUSION CRITERIA EXCLUSIONS OF PARTICIPANTS ELIGIBLE FOR ANALYSIS 330 Missing on binge drinking 92,670 Live-born singleton 80,781 1 st and 2 nd interview 6,002 1 st interview 86,783 Live-born singleton infants with data from 1 st interview 878 Missing on gestational age (GA) 4,441 > 1 week difference on own and register based GA 1,918 Missing on one or more covariates 330 Missing on binge drinking 878 Missing on GA 4,441 > 1 week difference on own and register based GA 765 Missing on one or more covariates 79,216 Paper IV 80,365 Paper V 100,418 Enrolled in DNBC Mar 18, 1996 Nov 1, ,043 Fetal death 443 Induced abortion 2,080 Multiple birth 5,887 No 1 st interview 86,783 Live-born singleton infants with data from 1 st interview 163 Missing on binge drinking 441 Missing on GA 2,228 > 1 week difference on own and register based GA 1381 Missing on one or more covariates 41,268 Paper VI 114 Hydatidiform mole/ectopic pregnancy 49 Emigrated 3 Deceased 16 Unknown

17 3.2. Measures of binge drinking and average alcohol intake The questions on the number and timing of episodes with binge drinking were identical in the first and second interviews (Table 1). Binge drinking was defined as an intake of five or more drinks on a single occasion. Information on when in pregnancy the women had binged was reported in commenced weeks (pregnancy weeks) calculated from the first day of the last menstrual cycle. The questions in the third interview were equivalent, but concerned the period from pregnancy week 30 and up to delivery. It should be emphasized that the respondents were unable to see the response categories and it was the interviewers who transferred the women s answers into the predefined response categories shown in Table 1. We used information from the first interview (or case interview in Paper III) to quantify binge drinking from onset of pregnancy to time of interview. In Papers IV to VI we retrieved the information on binge drinking from the second interview, if available, to quantify binge drinking in the period between the first and second interviews. Women were categorized as binge drinkers if they reported at least one episode of binge drinking. The number of binge episodes was categorized as 0, 1, 2, and 3+, except in Paper II in which it was categorized as 1, 2-3, and 4+ episodes. The timing of binge drinking was a part of the exposure of interest in Papers III to VI and categorized as yes vs. no in different developmental time periods of pregnancy. The division of pregnancy into different developmental time periods was fixed before the analyses and depended on the outcome of interest because the temporal sequence of developmental events is unique for each major organ system. 34 Thus, the division of pregnancy into developmental periods varied between the papers, please refer to Papers III to VI. In the first interview, the women were asked about their average weekly intake of beer, wine and spirits before pregnancy and during pregnancy, respectively (Table 1). One drink was defined as one bottle of beer, one glass of wine, or one glass of spirits, each of which corresponds to approximately 12 grams of alcohol in Denmark. 44 Women who reported to consume less than one drink per week of these alcoholic beverages were approximated to drink half a drink, and the weekly intake of the different alcoholic beverages was added up to a total. The categories of average intake were defined to ensure adequate numbers within categories and grouped into: non-drinkers, ½ -1½, 2-3½, and 4 or more drinks per week. Table 1. Questions from the first telephone interview on average alcohol intake and binge drinking wine is used as an example and similar questions were asked for beer and spirits as well as for the period before pregnancy How many glasses of wine do you drink per week now? 1. Number of glasses of wine 2. < 1 glass of wine/week 3. Never 4. Do not know 5. Do not want to answer If you think about the entire period of pregnancy also the very beginning how many times did you then have 5 drinks or more in one night/event? 1. No times 2. Number of times 3. Do not know 4. Do not want to answer What week(s) of gestation were you in the 1st, 2nd, 3rd etc. time? The interviews are available at 9

18 3.3. Handling incomplete information on timing of binge drinking Some of the women reporting binge drinking did not state the pregnancy week they were in for at least one of the episodes of binging. In Papers I and II these women were excluded, and in Paper III they were treated as unexposed if they did not state another binge episode in the developmental time period. In the data used in Papers IV to VI, around 8% of the binge drinkers did not state the timing for at least one of their binge episodes, and we imputed values of timing. A probability distribution of the timing of binge drinking was created from data on binge episodes with complete information on timing. This distribution was made dependent on whether the women recognized their pregnancy before pregnancy week 4, weeks 4-7, weeks 8-12, after week 12, or did not recall when they recognized their pregnancy. Information on time of interview or any other reported binge episodes was used to define a time interval of pregnancy in which every undated binge episode could have occurred. For example, if a woman reported two binge episodes and reported timing for the first episode, but did not recall the timing of the second, the second binge episode was assumed to have occurred between the timing of the first episode and the time of interview. To impute values for the missing information on timing we conditioned the probability distribution of timing to the possible time intervals. The imputation of missing information was made separately for the first and second interviews and data from the two interviews was combined afterwards. The imputation was performed five times and the presented estimates are the average of the estimates, while the standard errors used are based on both the variation within and between imputations Assessment of endpoints Papers III to VI are based on follow-up studies and the participants in the DNBC were followed in national registers to identify the pregnancy outcome, survival during the first year of life and presence of congenital anomalies (Table 2). In Denmark, every newborn citizen is assigned a unique civil registration number, which is a ten-digit number consisting of the date of birth and four digits. The civil registration number becomes inactive when the person dies. The women s civil registration numbers were used to identify all live births and stillbirths through linkage to the Civil Registration Register and the Medical Birth Register. At the time when the participants in the DNBC were pregnant, stillbirth was defined as birth of a dead fetus after 28 completed weeks of gestation, and losses of clinically recognized pregnancies before 28 completed weeks of gestation were recorded as spontaneous abortions. Pregnancy outcomes such as spontaneous abortion, early induced abortion, late induced abortion, hydatidiform mole and ectopic pregnancy were identified through linkage to the National Hospital Discharge Register. This register contains information on all diagnoses and surgical operations made on inpatients as well as outpatients. In less than 1% of pregnancies it was not possible to identify the pregnancy outcome through register linkage and information from the mothers was used to define the pregnancy outcome. Information on deaths during the first year of life and causes of death was obtained through linkage to the Civil Registration Register, the Medical Birth Register and the Register of Causes of Death. In the Register of Causes of Death data on deaths in 2002 and 2003 was temporary and the information in the register was insufficient for around 7% of deaths. Medical 10

19 records were obtained for these deaths in order to classify the cause of death. Information on congenital anomalies was obtained through register linkage to the National Hospital Discharge Register and the children were followed up to December 31,

20 Table 2. Outcomes in Papers III to VI and data source Paper and outcomes Population N Events N Data source* (end of follow up) Outcome definition (ICD-10 codes) Paper III 89,201 Fetal death 3,714 NHDR MBR Spontaneous abortions (O021, O028-O039) Induced abortions, fetal disease (O053, O054) Stillbirth (birth of death fetus 28 gestational weeks) Spontaneous abortion 3,270 Gestational age < 22 gestational weeks Stillbirth 444 Gestational age 22 gestational weeks Paper IV 79,216 Infant mortality Cause of death 279 CRR MBR RCD (The 1-year birth day) Death in the first year of life Aneuplodies (Q90-Q92) Congenital anomalies (Q00-Q89) Preterm birth (P072, P073) Pregnancy or delivery complications (P02, P03, P20, P21, P22) Sudden infant death (R95) Injuries (V43, V44, W75, W79, X94) Neonatal mortality 204 Death within the first 28 days of life Postneonatal mortality 75 Death on or after 28 days but within the first year of life Paper V 80, 365 Congenital heart disease 765 NHDR (Q20-Q26) Ventricular septal defects 279 (Dec. 31, 2006) (Q21.0) Atrial septal defects 258 (Q21.1) Paper VI 41,268 Cryptorchidism Mother report/diagnosis 1,598 Interview 3 or interview 4 Undescended testes at 6 or 18 months of age or diagnosis Diagnosis 810 NHDR (Q53, Q531, Q531A, Q532, Q532A, Q539) Diagnosis and orchiopexy 398 (Dec. 31, 2006) Surgical procedure KKFH00, KKFH01, KKFH10 *NHDR: National Hospital Discharge Register, MBR: Medical Birth Register, CRR: Civil Registration Register, RCD: Register of Causes of Death

21 3.5. Use of causal diagrams to select confounders The confounder selection, in all the papers on the health consequences of prenatal exposure to binge drinking, was based on causal diagrams in which we included prior knowledge from a literature search and logical reasoning about the causal relations between binge drinking, covariates and the health consequence of interest. These diagrams served as a way to visualize and explicitly elucidate the assumptions about the web of causation for the relation of interest and to identify variables to adjust for to obtain as unconfounded estimates of the causal effect between binge drinking and the health consequence as possible. As an example, the causal diagram we made to guide the confounder selection for the study on cryptorchidism is shown in Figure 3. The arrows represent the assumed causal relations between variables important to the relation between binge drinking and cryptorchidism. According to the diagram the analyses should be adjusted for maternal characteristics, smoking, time-to-pregnancy and diabetes. Gestational age at time of birth, other congenital malformations, fetal growth restriction and birth weight are possible intermediates on the pathway from binge drinking and cryptorchidism, and these were therefore not included in the analyses. Unfortunately we do not have data on use of illegal drugs, which according to the diagram is a potential confounder for the relation between binge drinking and cryptorchidism. The diagrams we have stated should not be taken as comprehensive causal models and some of our assumptions will definitely be open to debate, but by doing this exercise we clarified the assumptions that lied behind our statistical model Statistical methods Different types of statistical analyses were used in the different studies. In paper I we applied different measures of agreement, in Paper II data was analyzed by use of logistic regression analyses, in Papers III, IV and VI Cox proportional hazard models were used, and in Paper V the prevalence ratios of congenital heart disease, ventricular and atrial septal defects were estimated by log-binomial regression. The time axis in the Cox proportional hazard models was gestational age calculated as days since last menstrual period in the analyses of fetal death (Paper III) and child age in the analyses of survival during the first years of life and cryptorchidism. For more details on the statistical analyses please refer to Papers I to VI. Population attributable fractions were calculated for the increased risks in Papers III and IV, to estimate the proportion of all cases that would not have occurred if the women had not binged 3 or more times or consumed 4 or more drinks per week given causality. The population attributable fractions were estimated as: AF p p c RR 1 RR where pc is the exposure prevalence among cases and RR is the causal risk ratio. The estimated hazard ratios were used to estimate the causal risk ratios

22 Figure 3. Causal diagram for the relation between prenatal binge drinking during pregnancy and cryptorchidism Ethnicity Maternal characteristics: Age at birth Parity before this child Socio-occupational status Time to pregnancy Infertility treatment Smoking habits in pregnancy Personality affecting risk taking Use of illegal drugs Preclampsia Diabetes mellitus Binge drinking during pregnancy Cryptorchidism Fetal growth retardation Birth Weight Gestational age at birth/preterm delivery Twinning Congenital malformations Indicates that there exists no studies of this relation, but it is a part of my hypothesis Indicates that the variable is unmeasured

23 Proportion of women reporting binge drinking (%) 4. R E S U L T S 4.1. Repeatability of information on binge drinking (Paper I) When asked in mid-pregnancy (first interview), more women reported having binged at least once while being pregnant compared to when they were asked later in pregnancy (the second interview). The agreement between the reported information on binge drinking (yes versus no) in the first and second interviews was higher among women who answered the two interviews within 10 weeks compared to women with a longer time lag between the first and second interviews, kappa=0.63 versus kappa=0.54 (P< ). The number of binge drinking episodes was underestimated in the second interview when compared with the women s reported number of episodes in the first interview. Very few women reported binge drinking in pregnancy weeks 30 to 36; 84 women reported that they had binged when asked before giving birth, while only 37 women reported having binged in this period of pregnancy when asked again approximately six months after giving birth. A higher proportion of women reported that they had binged during the initial weeks of pregnancy when we compared the reports given in the first versus the second interview (Figure 4). The same was the case, when we compared women who answered the first interview early in pregnancy ( 12 week) versus later in pregnancy. Few women reported the same timing of their binge episodes in the first and second interviews. FIGURE 4. Proportion of women who reported binge drinking during the first 20 weeks of pregnancy, among women for whom information was obtained twice during pregnancy First interview Second interview Weeks of pregnancy 4.2. Characteristics of pregnant women who reported binge drinking (Paper II) Around one quarter of the women in the DNBC binged during pregnancy. Only 6% of women who reported binge drinking reported four or more episodes. Binge drinking was most common in the unrecognized part of pregnancy and only 3.5% of women reported binge drinking subsequent to recognition of pregnancy. Weekly alcohol consumption before pregnancy, single status and smoking, respectively, were associated with binge drinking both prior and subsequent to recognition of pregnancy. Generally, different 15

24 characteristics were associated with binge drinking in the unrecognized versus recognized part of pregnancy. Women who binged before becoming aware of their pregnancy were more often first-time mothers, aged years, in educated or skilled employment and had a time-to-pregnancy of less than one year. Whereas, women who binged once they became aware of their pregnancy were more likely to be unintended pregnant, multiparous, overweight or obese, have a mental disorder, be an unskilled worker, or unemployed for more than one year. Women who binged more than once after their pregnancy was confirmed were more often unintended pregnant, had been unemployed for more than one year, had a mental disorder, or consumed a higher amount of alcohol before pregnancy than women who binged only once in the recognized part of pregnancy Binge drinking in pregnancy and fetal death (Paper III) The number of binge episodes during the first 16 weeks of pregnancy was not associated with the risk of spontaneous abortion in clinically recognized pregnancies. The hazard ratios were similar for early ( 12 completed weeks) and late (13-21 completed weeks) spontaneous abortions. Women who binged in pregnancy weeks 11 to 16 had hazard ratios of 1.98 (95% CI: 0.97; 4.02) for late spontaneous abortion compared with women who did not binge during this period of pregnancy. Women who had binged three or more times during pregnancy had a hazard ratio of 1.56 (95% CI: 1.01; 2.40) for stillbirth compared with women without any binge episodes. Average alcohol consumption during pregnancy was positively associated with the risk of fetal death. Women who consumed three or more alcoholic drinks per week and binged at least twice during pregnancy had a hazard ratio of 2.20 (95% CI: 1.73; 2.80) for fetal death compared with women with no average intake and no binge drinking. When we restricted the analyses to women interviewed in pregnancy, in order to assess if recall bias influenced our findings, the positive association between average intake and fetal death was no longer present. It was not possible to clarify if the change in estimates was attributable to recall bias or differences in gestational age at fetal death for women who were interviewed during pregnancy versus subsequent to a fetal loss Alcohol drinking pattern and infant death (Paper IV) Children exposed to an average alcohol intake of at least four drinks per week had a hazard ratio for infant mortality of 1.78 (95% CI: 0.96; 3.32) compared with children who were unexposed to a weekly intake of alcohol. The increased risk was attributable to an increased hazard ratio of post-neonatal mortality of 3.56 (95% CI: 1.51; 8.43), while no increased hazard was seen for neonatal mortality. Children exposed to binge drinking three or more times during pregnancy had a hazard ratio of 2.69 (95% CI: 1.27; 5.69) for death during the post-neonatal period compared with children who were unexposed to binge drinking during pregnancy. The most common causes of death during the first year of life were congenital anomalies and preterm birth. The limited number of deaths during the first year of life prevented thorough analyses of separate causes of death. Among children born at term the hazard ratio of infant mortality was 2.71 (95% CI: 1.35; 5.45) for children exposed to an average intake of four or more drinks per week compared with unexposed children. Children who were born at term and prenatally exposed to three or 16

25 more times of binge drinking had hazard ratios of 1.97 (95% CI: 1.10; 3.54) for infant mortality compared with children who were unexposed to binge drinking. Post-neonatal mortality was responsible for the increase; the hazard ratio was 4.00 (95% CI: 1.55; 10.33) for an average weekly intake of four or more drinks and 2.93 (95% CI: 1.30; 6.57) for prenatal exposure to three or more binge episodes. Binge drinking during different developmental time periods of pregnancy was not associated with infant mortality, either in the neonatal or post-neonatal period of pregnancy Alcohol drinking pattern and congenital heart disease (Paper V) Prenatal exposure to alcohol on a weekly level, occasional binge drinking or binge drinking during the pre-conceptional period, the period of fertilization and implantation and/or organogenesis were not associated with an increased prevalence of congenital heart disease and the isolated forms ventricular and atrial septal defects. Findings on average intake may indicate an association with ventricular septal defect, but the test for trend was 0.23 and the prevalence ratio was 1.09 (95% CI: 0.96; 1.24) for every increase of one drink per week. The meager positive association between average intake and ventricular septal defects was of the same magnitude as an insignificant negative association between average intake and atrial septal defect Alcohol drinking pattern and cryptorchidism (Paper VI) The number of binge episodes and average intake during pregnancy were not associated with cryptorchidism. There were no statistically significant associations between binge drinking during different developmental stages of pregnancy and cryptorchidism. However, binge drinking in gestational weeks 7-15 was associated with a slightly increased risk of cryptorchidism, with hazard ratios between 1.03 and 1.66, but the confidence intervals were wide Population attributable fractions To estimate the percentages of stillbirths and deaths during the post-neonatal period that could be prevented if no women binged three or more times during pregnancy or consumed four or more drinks per week, we calculated the population attributable fraction. Based on the estimates on stillbirth, around 3% (95% CI: -0,3; 6) of the stillbirths in the DNBC could have been prevented if all the women who binged three or more times during pregnancy had not binged during pregnancy. If the women who binged three or more times during pregnancy had not exposed their fetuses to binge drinking 8% (95% CI: -1; 16) fewer post-neonatal deaths would have occurred. If they had binged less than three times 7% (95% CI: -1; 14) fewer post neonatal deaths would have occurred. Around 9% (95% CI: -1; 18) of deaths during the post-neonatal period in the DNBC could have been prevented if the women had abstained instead of consuming four or more alcoholic drinks per week. If we assume that these women decrease their intake to less than four drinks and not necessarily abstain from alcohol, around 6% (95% CI: -1; 12) of deaths during the post-neonatal period could have been prevented. This interpretation of the population attributable fraction as the potential caseload reduction assumes that all biases are absent, so that the observed association represents a valid estimate for the causal relation. The interpretation furthermore assumes that removing prenatal exposure 17

26 to binge drinking will not influence the size of the population at risk. This assumption may be unrealistic because alcohol consumption may affect the fecundity 131 and induce sub-clinical pregnancy loss, 132 and absence of alcohol would therefore probably increase the population at risk. 18

27 5. D I S C U S S I O N 5.1. Summary of main findings The findings presented in this thesis do not support the hypothesis that isolated episodes of binge drinking, in the absence of a consistently high daily intake, are associated with an increased risk of spontaneous abortion, congenital heart disease or cryptorchidism. Prenatal exposure to three or more episodes of binge drinking was associated with an increased risk of stillbirth and infant mortality, especially mortality in the post-neonatal period. A weekly intake of four or more drinks was also associated with an increased risk of infant mortality, primarily due to an increased risk of death during the post-neonatal period. Our findings do not provide any strong support for the hypothesis that the detrimental effects of binge drinking are time-specific and depend upon the fetus developmental stage at the time of binge drinking. However, the findings on cryptorchidism provide meager support for our hypothesis that binge drinking during weeks 8 to 14 could be of particular importance. Likewise, our findings indicate that binge drinking in weeks 11 to 16 may be related to late spontaneous abortion. Furthermore, our findings show that recall of binge drinking is affected by the length of the recall period, and women did not report the same timing of binge drinking when asked repeatedly. Finally, women who drink before versus after becoming aware of their pregnancy differ in social, life-style and reproductive characteristics. In the following sections, the findings of this thesis will be related to findings of other studies. Furthermore, the strengths and weaknesses of our studies will be discussed Consistency with previous studies Is it possible to obtain valid information on binge drinking during pregnancy? Maternal self-reporting of binge drinking during pregnancy provides the best possible estimation of such a drinking pattern, as no biomarkers exist that can definitively detect, quantify and establish timing of episodes with an increased blood alcohol level. 133 It has been suggested that it is possible to obtain information on the number of binge episodes as well as on the time of binge drinking by asking two simple questions similar to the questions used in the DNBC. 31 Validation of self-reports is crucial, but because no golden standard exists it is impossible to know what measure estimates closest to the true frequency and timing of binge drinking. Without information about the truth we are unable to address the validity, 134 and our findings represent the agreement between the answers obtained by use of the same questions at two different points in pregnancy. Since there is no reason to believe that pregnant women would systematically over-report their consumption of alcohol, it is a general approach to consider measures that state the highest reported intake to be the most valid When our findings are interpreted with this reasoning in mind, it implies that the first of two interviews referring to the same period probably provided the most valid measure of binge drinking, as more women reported binge drinking if the interview took place close to the period in question. These findings are supported by another study that also shows that the accuracy of recall diminishes with time as the reported average alcohol intake, even within a seven-day period, decreases with time. 138 The decline in accuracy of the reported information on 19

28 binge drinking as the recall period increases should be taken into consideration when planning or interpreting studies on the health consequences of binge drinking during pregnancy. For example, the major part of the existing studies on intrauterine exposure to binge drinking has collected information on binge drinking in the second part of pregnancy or subsequent to birth. 9;104; ;125;139. Therefore, the information on binge drinking is most likely underreported, and actually exposed women would have been categorized as non-exposed and the estimates may thereby have been biased toward the null. In the DNBC, the recall period is relatively short, because the women were interviewed early in pregnancy and interviewed twice during pregnancy. The information in the DNBC is therefore less likely to be biased due to underreporting than was the case in previous studies. Our finding of a low agreement on the reported timing of binge drinking is supported by another study that also reported a low week-by-week agreement. 40 Most of the disagreement between the two interviews on timing of binge drinking may be due to imprecise recollection of the true timing or inability to convert the calendar time of binge drinking into pregnancy time and report it in pregnancy weeks. The low agreement on the time in pregnancy when binging occurred is problematic, as it implies that the data on timing may be imprecise, and misclassification may have harmed our attempts to identify the periods of pregnancy in which the fetus is particularly vulnerable to binge drinking What characterize women in the DNBC who reported binge drinking? The prevalence of binge drinking, also among pregnant women, is likely to be country and time dependent as it varies according to the general alcohol consumption in a given country. 50 A number of studies have shown declining rates of alcohol consumption during pregnancy, e.g. in Norway and Denmark, over time. 48;140 In the United States, the consumption also decreased during the 1970s and until the beginning of the 1990s, but since then the consumption has been relatively steady The proportion of binge drinkers in the DNBC is substantially lower than in another survey of Danish pregnant women in which 50% reported binge drinking during pregnancy. 140 It is difficult to compare these two proportions because in neither of the surveys the participants were likely to be representative of all pregnant women in Denmark. The participants in the DNBC were on average healthier than the general population. 144 The other survey was conducted in one of the largest cities in Denmark and may therefore also not be representative of the country as a whole, as the proportion of binge drinkers has been shown to be lower in the more rural parts of Denmark. 145 Even though our study may be based on some of the healthiest pregnant women in Denmark, one in four reported binge drinking in pregnancy, and 3.5% of the women had one or more episode of binge drinking after recognition of pregnancy. Our study is the first to provide an estimate of binge drinking in the recognized part of pregnancy among Danish women. The proportion of pregnancy-aware binge drinkers is, however, similar to the approximately 4% of US women who are estimated to binge drink after recognition of pregnancy. 143 Also our findings on predictive factors of binge drinking during pregnancy are largely consistent with earlier findings, even though none of these studies distinguished between the pre- and post-recognized part of pregnancy. 31;48-50; Almost all of the existing 20

29 studies consistently report smoking as a predictor for binge drinking during pregnancy. 48;111;140;146;149;150 Alcohol consumption before pregnancy and single status have also previously been identified as predictive for binge drinking during pregnancy. 48;111;150 Nulliparous women are more likely to binge drink in the pre-recognized part of pregnancy than multiparous women. 140 This was the case for both intended and unintended pregnancies, which implies that pregnancy-planners do not modify their drinking habits before confirmation of pregnancy. This is consistent with the findings of other studies that have also reported drinking behavior to be unchanged until the pregnancy has been recognized However, once pregnancy was recognized, women without previous births were those least likely to report binge drinking. Our findings support that women experiencing an unintended pregnancy are more likely to binge than those who are intended pregnant. 152 Contrary to our findings, occupational status, socioeconomic position, and education have not been found to be associated with binge drinking during pregnancy in previous studies. 48;140;146;149;150 The existing results concerning age are contradictory. 111;149;150 In our study, the association between age and binge drinking was different for nulliparous and multiparous women as well as for binge drinking in the un- versus recognized part of pregnancy, and none of the previous studies have stratified the analyses according to these factors. It is interesting that binge drinking during pregnancy is not predicted by the same factors as the average intake during pregnancy. Two previous Danish studies have showed that the average intake during pregnancy is higher among older women, highly educated and employed women. 140;146 To summarize, binge drinking in the unrecognized part of pregnancy is quite common in Denmark, and in addition to smoking and single status, this drinking behavior is not more common among women who are unhealthy or socially disadvantaged Is prenatal exposure to binge drinking related to fetal death? Alcohol as an abortifacient has been the subject of many epidemiological studies, but very few have addressed the association between prenatal exposure to binge drinking and spontaneous abortion (Table 3). Similar to our findings, none of the previous studies have shown any association between prenatal exposure to binge drinking and spontaneous abortions. However, several studies have suggested an association between average intake and spontaneous abortion, primarily abortion during the first trimester. 4;5;7;8;132; Our findings, however, showed that binge drinking during weeks 11 to 16 was associated with an increased risk of late spontaneous abortion. This finding needs to be replicated in future studies, because no other studies have addressed the time-specific effect of binge drinking in relation to spontaneous abortion. Equally, no other study has investigated the potential association between binge drinking during pregnancy and stillbirth. The biological mechanisms explaining the association between prenatal alcohol exposure and spontaneous abortion are not established. However, findings from animal settings have indicated that high blood alcohol levels in early pregnancy may increase the incidence of embryonic resorption and spontaneous abortion. 29;34;71;159 Approximately 50-60% of all recognized spontaneous abortions have chromosome 21

30 abnormalities and approximately 70% of the abnormalities are aneuploidy. 160;161 Alcohol consumption around and shortly before conception may interfere the normal cell division and chromosome segregation, and thereby cause aneuploidy It is unclear whether it is ethanol per se, or its primary metabolite, acetaldehyde, that is the active substance, since both are capable of interfering with the functioning of the mitotic/meiotic spindle apparatus, and induces necrosis as well as congenital malformations. 160;161 The fact that the earliest cases of spontaneous abortion, those often caused by aneuploidy, are not captured in any prospective study in which the pregnancies are recruited once the pregnancy is recognized, may explain why our study and previous studies have not been able to show an association between binge drinking before and around conception and risk of spontaneous abortion. If the association between binge drinking during weeks 11 to 16 and late spontaneous abortion reflects a causal relation, it is plausible that this is caused by a direct lethal effect of binge drinking on the fetus. Given a priori knowledge on how alcohol may interfere with the normal cell division and the fact that none of the existing studies have been designed to study the very earliest cases of spontaneous abortion, it is still not possible to exclude the potential relation between binge drinking before and around conception and sub-clinical or very early pregnancy losses. On the other hand, frequency of binge drinking during the first 16 weeks of pregnancy may not have a great effect on spontaneous abortion in clinically recognized pregnancies, but binge drinking in weeks 11 to 16 may increase the risk of late spontaneous abortion. Further, our findings indicate that three or more episodes of binge drinking may increase the risk of stillbirth. 22

31 Table 3. Previous studies of binge drinking during pregnancy and spontaneous abortion Author Year Windham Cavallo Russel Jones Country Study design Study year USA Italy USA USA Cohort Cohort Not stated Cohort Cohort Not stated Population N (cases) Assessment of binge drinking Confounder control 5,342 (499) The greatest amount consumed on any one day since the last menstrual period. No information on frequency and timing of binge drinking 530 (58) Maximum daily intake during the last three months, categorized as 0, 1, 2+ drinks 244 (12) Categorical measure of episodes of having five or more drinks at a time 163 (Not stated) Binged 1 to 3 times during first trimester. Binged defined as to the point of feeling drunk Prior fetal loss, maternal race, education, age income, marital status, caffeine, smoking and employment status Age, marital status, occupation, parity, previous abortions, coffee and smoking Not stated Not stated Findings The greatest amount consumed on one occasion was not associated with an increased risk of spontaneous abortion. Among women who consumed > 3 drinks per week, the proportion of spontaneous abortion was greater if the women also reported 5 or more drinks at a single occasion than if she never drank that much in on day (31% vs. 21%) No association between maximum daily intake during the last three months and spontaneous abortion The results were not presented, but it was stated that this measure was not more significantly related to pregnancy outcomes than a continue measure for average absolute alcohol per day (ounces ~ 2 drinks) prior to pregnancy. OR=1.25 (1.02; 1.54) for spontaneous abortion for every ounce of alcohol consumed per day One to three times of binge drinking was not associated with risk spontaneous abortion. (The findings have never been published as a full paper)

32 Is maternal alcohol drinking patterns related to infant mortality? Like our study, other studies have indicated that intake of approximately five to six drinks per week is associated with an increased risk of infant mortality. 167;168 These previous studies did not sub-divide infant mortality into deaths during the neonatal vs. post-neonatal period. The previous studies on average intake and neonatal mortality also support our finding of no association. 117;169 Our study is the first to investigate the association between prenatal exposure to binge drinking and infant mortality, but a small case-control study among American Indians did support that maternal first-trimester binge drinking is a risk factor for the sudden infant death syndrome. 125 This syndrome is one of the leading causes of death during the post-neonatal period and might therefore be one of the causes of death accountable for the increased risk in the post-neonatal period. Furthermore, mechanisms involving alcohol s influence on brain areas involved in the cardiorespiratory control have been proposed to justify a causal relation between prenatal alcohol exposure and sudden infant death. 170 However, other studies have reported no association between prenatal alcohol exposure and the sudden infant death syndrome Preterm birth is one of the main contributors to death within the first year of life, especially during the first month of life. Maternal alcohol intake is associated with an increased risk of preterm delivery, and we therefore expected to find an association between maternal alcohol consumption and neonatal death. However, our findings suggested that the association with infant mortality was not mediated by an effect of alcohol on preterm delivery, as alcohol does not appear to have an effect on neonatal mortality nor on infant mortality among infants who were born preterm. The increased risk, especially during the post-neonatal period, may in addition to the sudden infant death syndrome be due to injuries or accidents. It is very plausible that women who consume alcohol during pregnancy will continue to drink and have more episodes of intoxication in the post-neonatal period, and children of these mothers may therefore be at an increased risk of injury-related death. Our study does not, however, support this explanation since handling the injury-related deaths as censored observations did not change our estimates. Prenatal alcohol exposure may limit the infant s ability to respond adequately to infections, 34;174 and infections may therefore also be a cause behind the association with infant death. Unfortunately, the limited number of deaths in our study prevented thorough analyses of causes of death. To summarize, our findings suggest that maternal drinking of at least four drinks per week or a minimum of three times of binge drinking are related to an increased risk of infant mortality, especially during the post-neonatal period. It was partly unexpected that the strongest associations were observed for post-neontal mortality, because neonatal deaths are much more common and alcohol have previously been linked to e.g. preterm birth, which is one of the main contributors to neonatal mortality Is prenatal exposure to alcohol related to congenital heart disease? The prevalence of congenital heart disease and especially septal defects is higher among infants of mothers with an alcohol use disorder than among infants who were not prenatally exposed to heavy alcohol drinking. 2;3; This implies that excessive alcohol 24

33 drinking or its correlates like poor diet, low social status etc. may be a cause of congenital heart disease. Findings from animal models support that alcohol exposure per se is at least partly responsible for this association because a relatively short exposure to high doses of alcohol has been demonstrated to cause congenital heart disease in mice and rats. 34;36;182;183 Additionally, findings from experimental research have suggested several tenable mechanisms between alcohol and abnormal heart development, including abnormal cell development, cell death, and reduction of the level of retinoic acid Several studies have examined the association between maternal alcohol consumption and the risk of congenital heart disease with somewhat conflicting findings ; Some smaller case-control studies have indicated an association between different levels of average intake and different forms of congenital heart disease, but in most of these studies the adjustment for confounding was very limited, if present at all ;187;188 The only previous studies that relied on prospectively collected data on alcohol intake support our finding of no association between average intake and congenital heart disease. 189;190 Few of the existing studies have included information on binge drinking (Table 4). Although, the definitions of binge drinking and of the time periods of pregnancy vary across the different studies, the results support our findings of no association between binge drinking and congenital heart disease, including septal defects. None of the existing studies included detailed information on timing of binge drinking, but all concerned exposure during the early part of pregnancy. To sum up, infants of women with an alcohol use disorder have a higher prevalence of congenital heart disease and especially septal defects, 175 but low-tomoderate levels of alcohol on a weekly basis or occasional binge drinking during the first trimester do not seem to cause congenital heart disease, ventricular or atrial septal defects Is there a time-specific effect of binge drinking on risk of cryptorchidism? Two prior studies have indicated an association between maternal alcohol consumption and cryptorchidism, but one of them showed an association only with transient cryptorchidism, 192 and the other was a small retrospective study that did not adjust for correlated behaviors such as smoking. 193 However, the majority of studies on maternal alcohol consumption in relation to cryptorchidism has shown no association In most of these studies, the information on alcohol was not very detailed, and only two studies included information on binge drinking. 192;200 Both of these studies found nonsignificantly increased odds ratios of cryptorchidism, but none of them included information about timing of binge drinking. 192;200 It has recently been hypothesized that the male fetus is particularly susceptible to androgen action around weeks 8 to 14 of gestation and deficient action during this period of pregnancy may program the androgen-dependent transinguinal testicular descent later in pregnancy. It is therefore plausible that the male fetus is particularly vulnerable to binge drinking during this period of pregnancy as alcohol has been shown to disrupt the hormone balance Our findings provide limited support to the hypothesis that the fetus is especially vulnerable to binge drinking during gestational weeks 8 to 14. However, it was intriguing that binge drinking during this period of pregnancy was associated with higher hazards of cryptorchidism, although insignificantly. In general, average alcohol intake and occasional binge drinking do not 25

34 seem to be risk factors for cryptorchidism. However, our findings provide some support for the hypothesis that binge drinking during gestational weeks 8 to 14 may increase the risk of cryptorchidism. 26

35 Table 4. Previous studies of binge drinking during pregnancy and congenital heart disease Author Year Grewal Wiliams Carmichael Tikkanen Country Study design Study year USA USA USA Case-control Case-control Case-control Finland Case-control Population N (cases) Outcome Assessment of binge drinking Confounder control Findings 700 (323) Conotruncal heart defects* 3029 (122) Isolated simple ventricular septal defects 481 (207) Conotruncal heart defects* 756 (50) Atrial septal defects Five or more drinks on one occasion during the first month after conception. Categorized as yes vs. no Five or more drinks on one occasion in the three months prior to pregnancy through the first trimester. Categorized as none, < once per week or once per week Five or more drinks on one occasion in the month before and the three months after conception. Categorized as never, < once per week or once per week At least 2-3 drinks per occasion during first trimester Maternal age, BMI, ethnicity, education, gravidity, employment, smoking and folic acid supplement use Maternal age, multivitamin use, diabetes and race Ethnicity, education, smoking and folic acid supplement use None No association No association No association with cronotruncal heart defects and teratogy of fallot. Binge drinking < once per week was associated with d- transposition of the great arteries, OR = 3.0 (95% CI: ) No association Tikkanen Finland Case-control (150) Ventricular septal defects More than 3 drinks per occasion during first trimester None No association Tikkanen Finland Case-control ,055 (573) Cardiovascular malformations More than 3 drinks per occasion during first trimester None No association Adams USA Case-control ,303 (83) Conotruncal heart defects* Six or more drinks on one occasion in the moth before conception through the 3 rd month of pregnancy. Categorized as yes vs. no None No association * Cronotruncal heart defects consisted of d-transposition of the great arteries, teratology of fallot, truncus arteriosus communis, double outlet right ventricle, aortico-pulmonary window, and subaortic ventricular septal defect type I in the study by Carmichael et al. Cases of the last mentioned were not included in the definition in the study by Adams et al and in the study by Grewal the definition only included cases of d-transposition of the great arteries and teratology of fallot.

36 5.3. Strengths and weaknesses The strengths of the Danish National Birth Cohort The DNBC is one of few maternal-child cohort studies that contains information on the number as well as on the timing of binge episodes and which is of sufficient size to separately address even rare pregnancy outcomes, including specific forms of congenital malformations. In addition, Denmark is one of few countries with a tradition of alcohol consumption during pregnancy and more than three quarters of pregnant women consider it acceptable to consume small amounts of alcohol while being pregnant. 52 Thus, the proportion of women who expose their pregnancy to alcohol on a weekly basis is higher than in many other countries. Likewise, weekend binge drinking is a common pattern of alcohol consumption. 210 Thus, the number of binge drinkers who otherwise drink little is assumed to be high, and confounding by the quantity of alcohol consumed is, therefore, assumed to be lower than in previous studies that are often based on both binge drinkers who otherwise drink little and binge drinkers who generally drink substantial amounts. 32 The predominately prospective design of the DNBC also ensured temporality between prenatal exposure to binge drinking and adverse pregnancy outcomes, which is important in studies of adverse pregnancy outcomes where recall problems may constitute a major problem. Linkage of civil registration numbers to nationwide population-based registers enabled identification of pregnancy outcomes for virtually all of the enrolled pregnancies and allowed nearly complete follow-up on the children. Despite the strengths of the DNBC, the observed associations may diverge from the causal relations by systematic processes, such as selection bias, misclassification of exposure or outcome, and confounding. The goal was to measure the relations as accurately as possible, but systematic errors in estimation are unavoidable and will therefore be considered in the following sections Selection bias Selection bias is present if the associations between prenatal exposure to binge drinking and adverse pregnancy outcome differ between the participants and those eligible for the study. 211 Selection bias may occur if the likelihood of being admitted to or followed in the sample depends on both the exposure and the outcome or bias may be more subtly related to factors that influence exposure and outcome. In cohort studies two sources of selection bias should be considered: incomplete follow up and non-participation in the study at time of enrollment into the cohort or at time of interview. Small proportions (less than 1%) of the women or children were lost to follow-up, and strong selection bias dur to loss to follow-up is therefore unlikely in Papers III and IV. In the studies on congenital anomalies the follow-up of the fetuses at risk is limited to live-born infants because anomalies in fetal losses and induced abortion are not recorded in national registries. This may induce bias, as it is equivalent to conditioning on prenatal survival. 129 If prenatal survival depended on both prenatal alcohol exposure and malformations, or shares a common cause with these factors, then conditioning the analysis to live-born infants would produce an estimate that is biased in a downward direction. 212 This type of bias may affect our findings on congenital heart disease (Figure 5), but probably not the 28

37 findings on cryptorchidism, as cryptorchidism is unlikely to have any strong impact on prenatal survival. The causal diagram in Figure 5 visualizes that a live-born infant is a common effect of prenatal alcohol exposure and congenital heart disease. By use of the simple graphical rules behind causal diagrams it is clear that conditioning on live-born infants will create a spurious association between prenatal alcohol exposure and congenital heart disease. Thus, the association between these two variables will not represent the causal relation, when the analyses are restricted to live-born infants, because a spurious association will distort the causal effect. This source of bias may not be a problem if the goal is to estimate the public health burden of prenatal alcohol exposure, however, the contrast of interest in etiological studies of malformations is the causal effect among all conceptuses. Figure 5. A causal diagram of loss to follow up when data are restricted to live-born infants and how this may bias the association between prenatal alcohol exposure and congenital heart disease Prenatal alcohol exposure Congenital heart disease Live-born infant Non-participation or self-selection into the DNBC may also be a source of bias. This bias may be of importance in all the included studies on health consequences. Only around 30% of the eligible population was actually enrolled into the DNBC and the participants were somewhat healthier than the eligible population. 144 Self-selection bias could be present because agreement to participate may be affected by the women s intake of alcohol and the women s past pregnancy experience. Thus, agreement to participate is a common effect of past pregnancy experience and alcohol consumption during pregnancy, as shown in the causal diagram in Figure 6. Past pregnancy experience can be used as a proxy for the risk of adverse outcome in the index pregnancy through genetic factors or exposures in past pregnancies that had a long-lasting effect. 213 Under these structures, the observed estimates will be biased when the analyses are restricted to those who volunteered to participate, because conditioning on agreement to participate will create a spurious association between past pregnancy experience and alcohol consumption during pregnancy. This source of selection bias may be reduced when the study base is restricted to women having no knowledge about their reproductive abilities, i.e. first-time pregnancies with a short time-to-pregnancy. We observed that this restriction influenced our findings on cryptorchidism and infant mortality. An inverse association was seen between average alcohol intake and cryptorchidism when the analyses were based on the entire cohort, whereas no association was seen when the data was restricted to first-time pregnancies with a time-to-pregnancy of less than 6 months. 29

38 On the other hand, the association between average intake and infant mortality was even more pronounced when data was restricted to women without any pregnancy experience. The same was seen for the association between the number of binge drinking episodes and infant mortality. Selection bias may also be present in Paper III, since only around two thirds of the women who had a spontaneous abortion before answering the first interview agreed to give a case interview, and we cannot exclude the possibility that there were women whose binge-like alcohol consumption influenced their willingness to give a case interview. Restricting the analyses to women who agreed to be interviewed could create a spurious association between binge drinking and fetal loss. However, in order to explain our findings the spurious association should be inverse and thereby might have biased an actual positive association between binge drinking and early fetal loss towards the null. Selection bias due to non-participation may constitute an even bigger issue in Paper II, where we addressed the characteristics of women who reported binge drinking, since it is possible that the decision to participate was influenced by both binge drinking and other of the identified characteristics. Figure 6. A causal diagram of the potential forces to self-selection into the Danish National Birth Cohort Alcohol consumption during pregnancy Adverse pregnancy outcome Agreement to participate Past pregnancy experience Misclassification of average alcohol intake and binge drinking Average intake and binge drinking may be misclassified, which would lead to significant bias. One of the major advantages of this study is that information on prenatal alcohol exposure was collected during pregnancy for all women, except the few who experienced an early pregnancy loss before they answered the scheduled interview. Thus, all information on alcohol was collected independently of the outcome except for some of the early cases of spontaneous abortion, and any misclassification of prenatal alcohol exposure is generally assumed to be non-differential. Women interviewed subsequent to fetal loss may have reported their alcohol intake differently from women who were still pregnant when interviewed, and this may lead to recall bias. 214 It is unlikely that our findings are due to recall bias as our estimates on binge drinking were similar when the analyses were restricted to those with data collected during pregnancy. However, recall bias may have influenced the association between average alcohol intake and 30

39 spontaneous abortion, but the change in estimates when restricting the analyses to those interviewed during pregnancy, could also be due to differences in gestational age at fetal death between women who were interviewed during pregnancy versus subsequent to a fetal loss. Another advantage of the self-reports in the DNBC is that the information was obtained by asking beverage-specific questions referring to a typical week during pregnancy, and by asking about the number as well as timing of each binge episode. It has been anticipated that more specific questions would result in a higher reported intake and thereby in a more valid estimate of the actual intake. 134 Likewise, interviews and questions referring to a typical week have been shown to be reliable methods to obtain information on alcohol consumption among pregnant Danish women. 135 The questions on binge drinking have also been suggested as a reliable method to obtain information on binge drinking. 31;40 However, no studies have ever assessed whether reporting of five or more drinks on a single occasion captures peaks in blood alcohol concentration, which is what we wanted to measure. An intake of five or more drinks will not result in the same blood alcohol concentration in every woman because of differences in e.g. body size. Some women will be intoxicated if they consume only 4 drinks. The women were not asked if they had been intoxicated, but it is likely that the women who report binge drinking had felt intoxicated to some extent, i.e. elevated alcohol concentration in the blood. Thus, it is likely that the women who were classified as exposed experienced an episode with increased blood alcohol concentration. Still, it is also likely that binge drinking is underreported and this may have biased our estimates toward the null. We showed that the recall of binge drinking was affected by the length of the recall period (Paper I), and therefore it is a strength of the data in the DNBC that the information on binge drinking was obtained twice during pregnancy as this shortens the recall time and thereby it is likely that binge drinking is underreported to a lesser degree. Timing of alcohol exposure is probably of importance, and our studies are the first to examine the effects of binge drinking during different developmental stages of pregnancy. However, we showed that the women did not state the same timing of binge drinking when asked repeatedly. The disagreement may be due to imprecise recollection of the true timing or inability to state the timing of episodes in pregnancy weeks. Irrespectively of why women were unable to state the same timing, it suggests that the information on timing may be imprecise and binge drinking during different developmental time periods may, therefore, be misclassified. Again, the information on timing was obtained in pregnancy and the misclassification is, therefore, most likely to be non-differential and this may explain why we were unable to identify time-specific effects of binge drinking. Furthermore, the potentially vulnerable periods may be of very short duration, i.e. maybe cover only a single day or two of the gestation, and the categorization of the different developmental time periods in our studies may have been too broad. Thus, we may have overlooked actual vulnerable time periods because we may have included binge episode outside the main period of susceptibility. 31 However, the data in the DNBC as well as knowledge about the exact vulnerable time period in humans are too imprecise to make a narrower categorization of pregnancy. 31

40 Misclassification of outcome measures The various endpoints are probably subject to some misclassification, i.e. the sensitivity and/or specificity is less than 100%. Any misclassification of outcome measure is thought to be non-differential, as data in the national registries is collected systematically and independently of any research. Furthermore, the identification of the given outcomes in the registries was made without knowing the exposure status of the women, thus any misclassification is likely to bias our findings toward the null. 214 Some stillbirths may have been given a civil registration number and thereby recorded as live-born infants who died on the day of delivery. Thus, the sensitivity may not be perfect. On the other hand, the specificity is assumed to be almost perfect as it is unlikely that a viable pregnancy that actually resulted in a live-born infant capable of surviving would have been recorded as a fetal death. In this scenario, non-differential misclassification is not affecting the hazard ratio. 212 The same reasoning applies for mortality during the first year of life and for boys who were diagnosed and operated for cryptorchidism as the specificity also is assumed to be almost perfect. However, for congenital heart disease it is impossible to rule out that a real effect may have been obscured by non-differential misclassification. The classification of all congenital heart diseases as one broad definition of outcome may also have diluted a real effect on specific forms of congenital heart disease. 212 Thus, our null finding on congenital heart defect may be due to misclassification bias Confounding In epidemiological studies in general, an unexposed group is used to provide an estimate of the disease risk among the exposed group, had they not been exposed. Confounding is present whenever the disease risk of the unexposed group differs from what it would have been in the exposed group had they not been exposed. 212 A covariate cannot be a confounder unless it can causally affect the disease risk within exposure groups, it is distributed differently among the compared groups, and it is not an intermediate variable in the causal pathway between exposure and disease. 218 In each of the four studies on health consequences we used causal diagrams to guide the confounder selection. The diagram we stated should not be taken as comprehensive causal models as they are simplified visual representations of our assumptions about the causal structures. Alternative assumptions can be justified, and in several of the studies sensitivity analyses were performed in order to address the impact of our assumptions. As an example, we assumed that birth weight, gestational age, and other malformations were possible intermediates on the pathway from prenantal alcohol exposure to cryptorchidism, and hence we did not adjust our analyses for these variables. It may be argued that these factors reflect the presence of confounders at baseline, which causes these outcomes, and in order to address the impact of these assumptions we performed sensitivity analyses where we adjusted for birth weight, gestational age and malformations. Such adjustments did not change the risk estimates. The fact that we did not have information (for all women) on the use of illegal drugs may be of especial concern, as it is likely that illegal drug use is more common among women who consume alcohol (especially high amounts) during pregnancy. 150 However, the use of illegal drugs during pregnancy is very low among 32

41 pregnant women in Denmark. 140;219 In the DNBC, information on illegal drug use was obtained through the second interview and only a very small percentage reported use of cannabis (0.2%) or other illegal drugs (0.05%). Furthermore, the relation between illegal drug use and cryptorchidism is not well established and strong confounding from illegal drug use is therefore unlikely. Other unmeasured confounding or residual confounding cannot be ruled out, as it is likely that women who consume the highest amounts of alcohol differ in a range of lifestyle factors from women who totally abstain during pregnancy. It is likely that alcohol consumption during pregnancy is influenced by past reproductive experience because of the widespread consensus that alcohol during pregnancy is harmful. 213 If past pregnancy experience is used to modify exposure status, we would expect women who consume alcohol to have a low a priori risk (as reflected by a successful pregnancy history in the past). Conversely, high-risk pregnancies are expected to be over-represented among the non-drinkers, if women with an unsuccessful pregnancy history or sub-fecundity take care to avoid hazardous exposures. Figure 7 reflects this scenario which resembles the structures from Figure 6, but a path from past pregnancy experience to alcohol consumption has been added. To alleviate this potential source of bias we restricted the analyses to a sub-cohort of women without any past pregnancy history to make use of when they decided to drink or to avoid alcohol in relation to their pregnancy. Our findings on cryptorchidism indicated an inverse association with average intake, but this may be due to behavior modification or selfselection into the cohort as no association was seen when analyzing the sub-cohort of first-time pregnancies conceived within six months of trying. Figure 7. A causal diagram of how past pregnancy experience may affect alcohol consumption during pregnancy and thereby bias the association between prenatal alcohol exposure and adverse pregnancy outcomes Alcohol consumption during pregnancy Adverse pregnancy outcome Agreement to participate Past pregnancy experience Random error Fortunately, stillbirth, infant death, especially post-neonatal death, and specific malformations such as cryptorchidism and spetal defects are rare. Even in a cohort as 33

42 large as the DNBC, the number of events is limited and it is not possible to exclude the possibility that causal relations may have been overlooked due to lack of statistical power. Furthermore, few women exposed their pregnancy to more than one episode of binge drinking or binged after recognition of pregnancy. For these reasons, the confidence intervals were wide and consequently, the findings regarding some of our outcomes should be interpreted with caution, even though the number of events in our studies is far higher than in previous studies. 34

43 6. C O N C L U S I O N A N D P E R S P E C T I V E S 6.1. Conclusion The findings presented in this thesis add substantially to the sparse knowledge about the potential health consequences of binge drinking during pregnancy. Exposure to three or more times of binge drinking during pregnancy was associated with an increased risk of stillbirth and infant mortality, primarily due to an increased risk of death during the post-neonatal period. The frequency of binge drinking was not associated with spontaneous abortion, congenital heart disease or cryptorchidism. The findings on cryptorchidism do, however, provide meager support for the hypothesis that weeks 8 to 14 may be a specific vulnerable time period for the testes development and binge drinking during this period of pregnancy may be of importance. Furthermore, binge drinking in weeks 11 to 16 was associated with an increased risk of late spontaneous abortion. As in any other epidemiological study, it is not possible to draw firm causal conclusions from the associations reported in our studies. For example, whether it is true that frequency of binge drinking is not causally related to spontaneous abortion, congenital heart disease and cryptorchidism, or whether selection bias or misclassification distorted a positive causal relation toward the null. Likewise, the positive associations between three or more episodes of binge drinking and stillbirth and infant death need not be causal, but could be attributable to unmeasured and/or residual confounding. Even though our studies are based on one of the largest maternal-child cohorts, and the number of events was therefore far larger than in previous studies, this ought to be kept in mind when interpreting some of our findings. Our findings were consistent with the limited prior scientific knowledge about the relations between binge drinking and these fetal health outcomes, and at the present time our findings constitute one of the best approximations to the causal relations. Furthermore, our studies were the first to address the associations between binge drinking and stillbirth and infant death, respectively. Therefore, our findings remain to be confirmed in other cohort studies. The outcomes we addressed only reflects a small part of fetal and child health and none of the addressed outcomes reflected neurodevelopmental issues, even though the brain development presumably is the organ most susceptible to the harmful effects of prenatal exposure to binge drinking. Nevertheless, the outcomes addressed in this thesis have previously been associated with prenatal alcohol exposure, and the susceptible time periods for these outcomes are expected to cover the earliest part of pregnancy, where binge drinking is most frequent Public health implications Exposure to binge drinking is largely a matter of personal choice, and the essential question for clinicians, public health specialists as well as for policy-makers who have the opportunity to intervene with preventive measures is: whether or not binge drinking is harmful in relation to pregnancy in women who otherwise drink nothing or low-tomoderate amounts? In order to answer this question the range of outcomes is wide and varied, e.g. areas deserving attention range from fecundity to disorders appearing in 35

44 adult life of the child-to-be, such as reduced sperm counts or specific malignancies. 7;220 Consequently, it is obvious that the possible adverse outcomes are many, and we have to decide how many outcomes should be examined before we are convinced that binge drinking during pregnancy causes no harm. Whenever different levels of exposure are associated with more than one outcome, then the issue of establishing a public health policy (or even more; a safe level of exposure) becomes complex. In these instances it is often necessary to balance positive and negative findings. 221 Further, it may be necessary to decide if advice is to be given in relation to the outcome with the most severe health consequences or the outcome, which is sensitive to the lowest level of exposure. 7 At the present time, relatively few epidemiological studies have examined the association between binge drinking and fetal and child health and almost nothing is known about the long-term implications that will appear during adult life. 32;103 When synthesizing the sparse evidence from humans, it seems no consistent evidence exists of adverse effects of occasional binge drinking during pregnancy, despite a possible effect on neurodevelopmental outcomes. The limits of observational epidemiology, such as insufficient power, selection bias and misclassification may also have biased previous studies, as well as our studies, toward the null and true hazards may have been overlooked. However, we showed that frequent intrauterine exposure to binge drinking may influence survival, and in this thesis it was shown that 8% of post-neonatal deaths and 3% of stillbirths in our studies could have been prevented if the women had not binged during pregnancy. Moreover, knowledge about the harmful effects in animals provides us with plausible and credible evidence of likely and substantial harm in humans from intrauterine exposure to binge drinking. Thus, from a public health point of view, and based on the precautionary principle, 222;223 it may be worthwhile recommending pregnant women and women who are planning a pregnancy to avoid binge drinking. It may also be worthwhile to include this information explicitly in the countries that recommend total abstinence during pregnancy, because binge drinking is presumed to be even more harmful than consumption of the same amount or even a higher amount over a longer duration of time. This may prevent pregnant women, who in spite of warnings against drinking during pregnancy decide to drink, from lumping together a given amount of alcohol to a few events in the presumption that this is less hazardous than a frequent low alcohol intake. A further argument in favor of the precautionary principle is that binge drinking is not presumed to be healthy for anything e.g. not even coronary heart disease and it has been shown to be associated with increased all cause mortality. 224;225 For that reason, the Danish guidelines for sensible drinking also explicitly include binge drinking, and men as well as women are advised not to drink more than five drinks in any session. 221 Thus, it seems reasonable for an official public health policy to recommend that binge drinking should be avoided in relation to pregnancy. However, equally from a clinician s point of view, the findings from this thesis may also be used to reassure pregnant women who worry about having binged, because the clinicians can inform the women about the minimal risk of adverse pregnancy outcomes associated with isolated episodes of binge drinking in the absence of a consistently high daily alcohol intake. 36

45 7. S U M M A R Y I N E N G L I S H Binge drinking, which is usually defined as consumption of five or more drinks per sitting, is a common way to consume alcohol in Denmark and other Western countries. This drinking pattern is especially widespread among young people and may, therefore, also occur among women with an unrecognized pregnancy. It has been reported from Norway and Denmark that 25 and 50%, respectively, of women have binged during pregnancy. Eighty-five percent of Danish women consider binge drinking to be harmful to the fetus, and binge drinking in the unrecognized part of pregnancy is often a source of concern. Health professionals involved in antenatal care programs are therefore often consulted about possible health effects of isolated episodes of binge drinking in early pregnancy. The lack of strong evidence from epidemiological studies makes it difficult to advise these women. The aim of this PhD thesis is to address the health consequences of prenatal exposure to binge drinking. Studies included in the thesis are all based on data from the Danish National Birth Cohort (DNBC), a population-based birth cohort of around 100,000 pregnancies with information on the number and timing of binge drinking in pregnancy. Initially, we examined women s reporting of binge drinking and further characterized the women who had binged during pregnancy. Around one quarter of the women reported binge drinking, in most cases before they became aware of their pregnancy. A higher proportion of women reported binge drinking, if the interview took place close to the period in question than if asked later on. A weekly alcohol consumption before pregnancy, smoking habits and single status were correlated with binge drinking both in the unrecognized and recognized parts of pregnancy. Moreover, binge drinking in the pre-recognized part of pregnancy was more common among women who were primiparous, in their late twenties, educated or in good jobs. Binge drinking in the recognized part of pregnancy was more common among women who had become unintended pregnant, were multiparous, had a mental disorder, were in unskilled jobs or had been unemployed for more than one year. Pregnancies that were exposed to three or more binge episodes were more likely to end in stillbirth, or in the infant s death within the first year of life, especially during the postneonatal period. The number of binge episodes was not associated with risk of spontaneous abortion, congenital heart disease or cryptorchidism. Our findings bring limited support to the hypothesis that binge drinking during gestational weeks 8 to 14 may be important for the testes descent later in pregnancy, as binge drinking during this period of pregnancy was associated with increased risk of cryptorchidism, although insignificant. Furthermore, binge drinking in weeks 11 to 16 was associated with an increased risk of late spontaneous abortion. The findings from this thesis indicate that one or two episodes of binge drinking during the early part of pregnancy are not associated with an increased risk of spontaneous abortion, stillbirth, infant mortality, congenital heart disease or cryptorchidism. These findings only reflect a part of the potential health consequences of prenatal exposure to binge drinking and data from the DNBC constitute an excellent opportunity for further research. 37

46 8. S U M M A R Y I N D A N I S H Episodisk højt alkoholindtag, binge drinking, som ofte er defineret som indtag af fem genstande eller derover ved en enkelt lejlighed, forekommer hyppigt i Danmark og andre vestlige lande og er specielt udbredt blandt unge. Binge drinking forekommer derfor også hyppigt blandt kvinder, der er uerkendt gravide. Tal fra Norge og Danmark viser, at hhv. 25 og 50% af gravide har binge drukket i løbet af deres graviditet. Hovedparten af gravide i Danmark (85%) anser binge drinking under graviditeten for at være potentielt skadeligt. Derfor er mange gravide, der har binge drukket inden de var klar over, at de var gravide, bekymrede for, om deres foster har taget skade. Det er almindeligt, at de læger og jordmødre der er involveret i svangeromsorgen, bliver spurgt til helbredskonsekvenserne af at have binge drukket få gange helt tidligt i graviditeten. Dette spørgsmål er vanskeligt at besvare, idet evidensen fra epidemiologiske undersøgelser er sparsom. Formålet med denne ph.d.-afhandling er at undersøge mulige føtale helbredskonsekvenser af prænatal eksponering for binge drinking. Samtlige undersøgelser i afhandlingen er baseret på data fra den nationale fødselskohorte Bedre Sundhed for Mor og Barn, hvori der er indsamlet detaljerede data, på ca. 100,000 graviditeter, om antal af og tidspunkt for binge drinking. Indledningsvis undersøgte vi de gravides rapportering af binge drinking og dernæst, hvad der karakteriserede de kvinder, som binge drak under graviditeten. I denne kohorte af gravide havde ca. hver fjerde binge drukket under graviditeten, som oftest inden graviditeten var erkendt. Flere kvinder rapporterede binge drinking, hvis de blev interviewet kort vs. lang tid efter den periode de skulle angive binge drinking for. Binge drinking både i den uerkendte og erkendte del af graviditeten forekom hyppigere blandt kvinder, der drak alkohol før de blev gravide, røg og ikke boede sammen med det kommende barns far. Binge drinking i den uerkendte del af graviditeten forekom derudover hyppigere blandt førstegangsgravide, der var sidst i tyverne og havde en uddannelse eller var i gode jobs. Gravide, der binge drak efter erkendelse af graviditeten var derimod oftere uønsket gravide, flergangsfødende, havde en psykisk lidelse, var ufaglærte eller havde været udenfor arbejdsmarkedet i mere end et år. Kvinder der rapporterede tre eller flere episoder med binge drinking havde en øget risiko for dødfødsel og spædbarnsdød, specielt post-neonatal død. Antal af episoder med binge drinking under graviditeten var ikke associeret til risikoen for spontan abort, medfødte hjertemisdannelser eller kryptorkisme. Vores fund understøtter svagt hypotesen om, at binge drinking i løbet af gestationsuge 8 til 14 er vigtigt for at testiklerne synker ned i pungen senere i graviditeten, idet binge drinking i løbet af denne periode af graviditeten, var associeret med forøget risiko for kryptorkisme, resultaterne var dog ikke statistisk signifikante. Derudover var binge drinking i uge 11 til 16 associeret med en forøget risiko for spontan abort i uge 13 eller derefter. Resultaterne fra denne afhandling tyder på at en eller to lejligheder med binge drinking i den tidlige graviditet ikke forøger risikoen for spontan abort, dødfødsel, spædbarnsdød, hjertemisdannelser eller kryptorksime. Disse udfald udgør kun en lille del af potentielle helbredseffekter, og data fra Bedre Sundhed for Mor og Barn vil være ideelt til undersøgelse af andre mulige helbredseffekter. 38

47 9. R E F E R E N C E S (1) Jones KL, Smith DW, Ulleland CN, Streissguth P. Pattern of malformation in offspring of chronic alcoholic mothers. Lancet 1973; 1(7815): (2) Jones KL, Smith DW. Recognition of the fetal alcohol syndrome in early infancy. Lancet 1973; 2(7836): (3) Lemoine P, Harousseau H, Borteyru JP, Menuet JC. Children of alcoholic parents- -observed anomalies: discussion of 127 cases. Ther Drug Monit 2003; 25(2): (4) Windham GC, Von Behren J, Fenster L, Schaefer C, Swan SH. Moderate maternal alcohol consumption and risk of spontaneous abortion. Epidemiology 1997; 8(5): (5) Windham GC, Fenster L, Swan SH. Moderate maternal and paternal alcohol consumption and the risk of spontaneous abortion. Epidemiology 1992; 3(4): (6) Harlap S, Shiono PH. Alcohol, smoking, and incidence of spontaneous abortions in the first and second trimester. Lancet 1980; 2(8187): (7) Kline J, Levin B, Stein Z, Susser M, Warburton D. Epidemiologic detection of low dose effects on the developing fetus. Environ Health Perspect 1981; 42: (8) Kesmodel U, Wisborg K, Olsen SF, Henriksen TB, Secher NJ. Moderate alcohol intake in pregnancy and the risk of spontaneous abortion. Alcohol Alcohol 2002; 37(1): (9) Passaro KT, Little RE, Savitz DA, Noss J. The effect of maternal drinking before conception and in early pregnancy on infant birthweight. The ALSPAC Study Team. Avon Longitudinal Study of Pregnancy and Childhood. Epidemiology 1996; 7(4): (10) Lundsberg LS, Bracken MB, Saftlas AF. Low-to-moderate gestational alcohol use and intrauterine growth retardation, low birthweight, and preterm delivery. Ann Epidemiol 1997; 7(7): (11) Kesmodel U, Olsen SF, Secher NJ. Does alcohol increase the risk of preterm delivery? Epidemiology 2000; 11(5): (12) Albertsen K, Andersen AM, Olsen J, Gronbaek M. Alcohol consumption during pregnancy and the risk of preterm delivery. Am J Epidemiol 2004; 159(2):

48 (13) Nanson JL, Hiscock M. Attention deficits in children exposed to alcohol prenatally. Alcohol Clin Exp Res 1990; 14(5): (14) Korkman M, Hilakivi-Clarke LA, utti-ramo I, Fellman V, Granstrom ML. Cognitive impairments at two years of age after prenatal alcohol exposure or perinatal asphyxia. Neuropediatrics 1994; 25(2): (15) Streissguth AP, Barr HM, Sampson PD, Parrish-Johnson JC, Kirchner GL, Martin DC. Attention, distraction and reaction time at age 7 years and prenatal alcohol exposure. Neurobehav Toxicol Teratol 1986; 8(6): (16) Streissguth AP, Sampson PD, Barr HM. Neurobehavioral dose-response effects of prenatal alcohol exposure in humans from infancy to adulthood. Ann N Y Acad Sci 1989; 562: (17) Streissguth AP, Barr HM, Sampson PD. Moderate prenatal alcohol exposure: effects on child IQ and learning problems at age 7 1/2 years. Alcohol Clin Exp Res 1990; 14(5): (18) Kelly Y, Sacker A, Gray R, Kelly J, Wolke D, Quigley MA. Light drinking in pregnancy, a risk for behavioural problems and cognitive deficits at 3 years of age? Int J Epidemiol (19) O'Brien P. Is it all right for women to drink small amounts of alcohol in pregnancy? Yes. BMJ 2007; 335(7625):856. (20) Nathanson V, Jayesinghe N, Roycroft G. Is it all right for women to drink small amounts of alcohol in pregnancy? No. BMJ 2007; 335(7625):857. (21) Mukherjee RA, Hollins S, Abou-Saleh MT, Turk J. Low level alcohol consumption and the fetus. BMJ 2005; 330(7488): (22) Henderson J, Gray R, Brocklehurst P. Systematic review of effects of lowmoderate prenatal alcohol exposure on pregnancy outcome. BJOG 2007; 114(3): (23) Pierce DR, West JR. Blood alcohol concentration: a critical factor for producing fetal alcohol effects. Alcohol 1986; 3(4): (24) Bonthius DJ, West JR. Blood alcohol concentration and microencephaly: a doseresponse study in the neonatal rat. Teratology 1988; 37(3): (25) West JR, Goodlett CR, Bonthius DJ, Hamre KM, Marcussen BL. Cell population depletion associated with fetal alcohol brain damage: mechanisms of BACdependent cell loss. Alcohol Clin Exp Res 1990; 14(6):

49 (26) Goodlett CR, Eilers AT. Alcohol-induced Purkinje cell loss with a single binge exposure in neonatal rats: a stereological study of temporal windows of vulnerability. Alcohol Clin Exp Res 1997; 21(4): (27) Clarren SK, Astley SJ, Gunderson VM, Spellman D. Cognitive and behavioral deficits in nonhuman primates associated with very early embryonic binge exposures to ethanol. J Pediatr 1992; 121(5 Pt 1): (28) Clarren SK, Astley SJ, Bowden DM. Physical anomalies and developmental delays in nonhuman primate infants exposed to weekly doses of ethanol during gestation. Teratology 1988; 37(6): (29) Stuckey E, Berry CL. The effects of high dose sporadic (binge) alcohol intake in mice. J Pathol 1984; 142(3): (30) Gladstone J, Nulman I, Koren G. Reproductive risks of binge drinking during pregnancy. Reprod Toxicol 1996; 10(1):3-13. (31) Kesmodel U. Binge drinking in pregnancy--frequency and methodology. Am J Epidemiol 2001; 154(8): (32) Henderson J, Kesmodel U, Gray R. Systematic review of the fetal effects of prenatal binge-drinking. J Epidemiol Community Health 2007; 61(12): (33) Shepard TH, Brent RL, Friedman JM, Jones KL, Miller RK, Moore CA et al. Update on new developments in the study of human teratogens. Teratology 2002; 65(4): (34) Becker HC, Diaz-Granados JL, Randall CL. Teratogenic actions of ethanol in the mouse: a minireview. Pharmacol Biochem Behav 1996; 55(4): (35) Cavieres MF, Smith SM. Genetic and developmental modulation of cardiac deficits in prenatal alcohol exposure. Alcohol Clin Exp Res 2000; 24(1): (36) Daft PA, Johnston MC, Sulik KK. Abnormal heart and great vessel development following acute ethanol exposure in mice. Teratology 1986; 33(1): (37) Maier SE, Miller JA, Blackwell JM, West JR. Fetal alcohol exposure and temporal vulnerability: regional differences in cell loss as a function of the timing of bingelike alcohol exposure during brain development. Alcohol Clin Exp Res 1999; 23(4): (38) Thomas JD, Wasserman EA, West JR, Goodlett CR. Behavioral deficits induced by bingelike exposure to alcohol in neonatal rats: importance of developmental timing and number of episodes. Dev Psychobiol 1996; 29(5):

50 (39) Allebeck P, Olsen J. Alcohol and fetal damage. Alcohol Clin Exp Res 1998; 22(7 Suppl):329S-332S. (40) Kesmodel U, Frydenberg M. Binge Drinking during Pregnancy-Is It Possible to Obtain Valid Information on a Weekly Basis? Am J Epidemiol 2004; 159(8): (41) Department of health. The pregnancy book. London: The Stationery Office; (42) Kesmodel U. [Alcohol and pregnancy]. Ugeskr Laeger 1999; 161(36): (43) Sundhedsstyrelsen. Alkohol og graviditet. Sundhedsstyrelsens hjemmeside, available in Danish at: (44) Sundhedsstyrelsen. Graviditet & alkohol - Forebyggelse og sundhedsfremme nr.15, København: Sundhedsstyrelsen; (45) Tough S, Tofflemire K, Clarke M, Newburn-Cook C. Do women change their drinking behaviors while trying to conceive? An opportunity for preconception counseling. Clin Med Res 2006; 4(2): (46) Kost K, Landry DJ, Darroch JE. Predicting maternal behaviors during pregnancy: does intention status matter? Fam Plann Perspect 1998; 30(2): (47) Hellerstedt WL, Pirie PL, Lando HA, Curry SJ, McBride CM, Grothaus LC et al. Differences in preconceptional and prenatal behaviors in women with intended and unintended pregnancies. Am J Public Health 1998; 88(4): (48) Alvik A, Heyerdahl S, Haldorsen T, Lindemann R. Alcohol use before and during pregnancy: a population-based study. Acta Obstet Gynecol Scand 2006; 85(11): (49) Naimi TS, Brewer RD, Mokdad A, Denny C, Serdula MK, Marks JS. Binge drinking among US adults. JAMA 2003; 289(1): (50) Kuntsche E, Rehm J, Gmel G. Characteristics of binge drinkers in Europe. Soc Sci Med 2004; 59(1): (51) BMA Board of science. Fetal alcohol spectrum disorders - A guide for health care professionals. British Medical Association; (52) Kesmodel U, Schioler KP. Drinking during pregnancy: attitudes and knowledge among pregnant Danish women, Alcohol Clin Exp Res 2002; 26(10):

51 (53) Enkin M, Keirse MJNC, Neilson J, Crowther C, Duley L, Hodnett E et al. A guide to effective care in pregnancy and childbirth. 3rd ed. New York: Oxford University Press; (54) Idanpaan-Heikkila J, Jouppila P, Akerblom HK, Isoaho R, Kauppila E, Koivisto M. Elimination and metabolic effects of ethanol in mother, fetus, and newborn infant. Am J Obstet Gynecol 1972; 112(3): (55) Mann LI, Bhakthavathsalan A, Liu M, Makowski P. Placental transport of alcohol and its effect on maternal and fetal acid-base balance. Am J Obstet Gynecol 1975; 122(7): (56) Burd L, Roberts D, Olson M, Odendaal H. Ethanol and the placenta: A review. J Matern Fetal Neonatal Med 2007; 20(5): (57) Michaelis EK, Michaelis ML. Cellular and Molecular Bases of Alcohol's Teratogenic Effects. Alcohol Health and Research World 1994; 18: (58) Hard ML, Einarson TR, Koren G. The role of acetaldehyde in pregnancy outcome after prenatal alcohol exposure. Ther Drug Monit 2001; 23(4): (59) Lee RD, An SM, Kim SS, Rhee GS, Kwack SJ, Seok JH et al. Neurotoxic effects of alcohol and acetaldehyde during embryonic development. J Toxicol Environ Health A 2005; 68(23-24): (60) Pratt OE. Introduction: what do we know of the mechanisms of alcohol damage in utero? Ciba Found Symp 1984; 105:1-7. (61) Goodlett CR, Horn KH. Mechanisms of alcohol-induced damage to the developing nervous system. Alcohol Res Health 2001; 25(3): (62) Goodlett CR, Horn KH, Zhou FC. Alcohol teratogenesis: mechanisms of damage and strategies for intervention. Exp Biol Med (Maywood ) 2005; 230(6): (63) West JR, Goodlett CR, Bonthius DJ, Pierce DR. Manipulating peak blood alcohol concentrations in neonatal rats: review of an animal model for alcohol-related developmental effects. Neurotoxicology 1989; 10(3): (64) Bonthius DJ, Goodlett CR, West JR. Blood alcohol concentration and severity of microencephaly in neonatal rats depend on the pattern of alcohol administration. Alcohol 1988; 5(3): (65) Bonthius DJ, West JR. Alcohol-induced neuronal loss in developing rats: increased brain damage with binge exposure. Alcohol Clin Exp Res 1990; 14(1):

52 (66) Goodlett CR, Thomas JD, West JR. Long-term deficits in cerebellar growth and rotarod performance of rats following "binge-like" alcohol exposure during the neonatal brain growth spurt. Neurotoxicol Teratol 1991; 13(1): (67) Maier SE, West JR. Drinking patterns and alcohol-related birth defects. Alcohol Res Health 2001; 25(3): (68) Ramchandani VA, Bosron WF, Li TK. Research advances in ethanol metabolism. Pathol Biol (Paris ) 2001; 49(9): (69) Chen WJ, Parnell SE, West JR. Effects of alcohol and nicotine on developing olfactory bulb: loss of mitral cells and alterations in neurotransmitter levels. Alcohol Clin Exp Res 1999; 23(1): (70) Barnard J, Rubin DR, Schenker N. Multiple imputation methods. In: Armitage P, Colton T, editors. Encyclopedia of Biostatistics. John Wiley & Sons; (71) Clarren SK, Bowden DM, Astley SJ. Pregnancy outcomes after weekly oral administration of ethanol during gestation in the pig-tailed macaque (Macaca nemestrina). Teratology 1987; 35(3): (72) Padmanabhan R, Muawad WM. Exencephaly and axial skeletal dysmorphogenesis induced by acute doses of ethanol in mouse fetuses. Drug Alcohol Depend 1985; 16(3): (73) Padmanabhan R, Hameed MS, Sugathan TN. Effects of acute doses of ethanol on pre- and postnatal development in the mouse. Drug Alcohol Depend 1984; 14(2): (74) Sulik KK, Johnston MC, Webb MA. Fetal alcohol syndrome: embryogenesis in a mouse model. Science 1981; 214(4523): (75) Goodlett CR, Marcussen BL, West JR. A single day of alcohol exposure during the brain growth spurt induces brain weight restriction and cerebellar Purkinje cell loss. Alcohol 1990; 7(2): (76) Goodlett CR, Mahoney JC, West JR. Brain growth deficits following a single day of alcohol exposure in the neonatal rat. Alcohol 1989; 6(2): (77) West JR, Kelly SJ, Pierce DR. Severity of alcohol-induced deficits in rats during the third trimester equivalent is determined by the pattern of exposure. Alcohol Alcohol 1987; Suppl 1: (78) Ikonomidou C, Bittigau P, Ishimaru MJ, Wozniak DF, Koch C, Genz K et al. Ethanol-induced apoptotic neurodegeneration and fetal alcohol syndrome. Science 2000; 287(5455):

53 (79) Maier SE, Miller JA, West JR. Prenatal binge-like alcohol exposure in the rat results in region-specific deficits in brain growth. Neurotoxicol Teratol 1999; 21(3): (80) Clarren SK, Bowden DM. Fetal alcohol syndrome: a new primate model for binge drinking and its relevance to human ethanol teratogenesis. J Pediatr 1982; 101(5): (81) Sheller B, Clarren SK, Astley SJ, Sampson PD. Morphometric analysis of Macaca nemestrina exposed to ethanol during gestation. Teratology 1988; 38(5): (82) Clarren SK, Astley SJ, Bowden DM, Lai H, Milam AH, Rudeen PK et al. Neuroanatomic and neurochemical abnormalities in nonhuman primate infants exposed to weekly doses of ethanol during gestation. Alcohol Clin Exp Res 1990; 14(5): (83) Clarren SK, Astley SJ. Pregnancy outcomes after weekly oral administration of ethanol during gestation in the pig-tailed macaque: comparing early gestational exposure to full gestational exposure. Teratology 1992; 45(1):1-9. (84) Bonthius DJ, Bonthius NE, Napper RM, Astley SJ, Clarren SK, West JR. Purkinje cell deficits in nonhuman primates following weekly exposure to ethanol during gestation. Teratology 1996; 53(4): (85) Miller MW, Astley SJ, Clarren SK. Number of axons in the corpus callosum of the Mature macaca nemestrina: increases caused by prenatal exposure to ethanol. J Comp Neurol 1999; 412(1): (86) Astley SJ, Magnuson SI, Omnell LM, Clarren SK. Fetal alcohol syndrome: changes in craniofacial form with age, cognition, and timing of ethanol exposure in the macaque. Teratology 1999; 59(3): (87) West JR, Parnell SE, Chen WJ, Cudd TA. Alcohol-mediated Purkinje cell loss in the absence of hypoxemia during the third trimester in an ovine model system. Alcohol Clin Exp Res 2001; 25(7): (88) Ramadoss J, Lunde ER, Pina KB, Chen WJ, Cudd TA. All three trimester binge alcohol exposure causes fetal cerebellar purkinje cell loss in the presence of maternal hypercapnea, acidemia, and normoxemia: ovine model. Alcohol Clin Exp Res 2007; 31(7): (89) Ramadoss J, Hogan HA, Given JC, West JR, Cudd TA. Binge alcohol exposure during all three trimesters alters bone strength and growth in fetal sheep. Alcohol 2006; 38(3): (90) Parnell SE, Ramadoss J, Delp MD, Ramsey MW, Chen WJ, West JR et al. Chronic ethanol increases fetal cerebral blood flow specific to the ethanol-sensitive 45

54 cerebellum under normoxaemic, hypercapnic and acidaemic conditions: ovine model. Exp Physiol 2007; 92(5): (91) Ramadoss J, Lunde ER, Chen WJ, West JR, Cudd TA. Temporal vulnerability of fetal cerebellar Purkinje cells to chronic binge alcohol exposure: ovine model. Alcohol Clin Exp Res 2007; 31(10): (92) Zajac CS, Abel EL. Animal models of prenatal alcohol exposure. Int J Epidemiol 1992; 21 Suppl 1:S24-S32. (93) Nulman I, Rovet J, Kennedy D, Wasson C, Gladstone J, Fried S et al. Binge alcohol consumption by non-alcohol-dependent women during pregnancy affects child behaviour, but not general intellectual functioning; a prospective controlled study. Arch Women Ment Health 2004; 7(3): (94) Bailey BN, Delaney-Black V, Covington CY, Ager J, Janisse J, Hannigan JH et al. Prenatal exposure to binge drinking and cognitive and behavioral outcomes at age 7 years. Am J Obstet Gynecol 2004; 191(3): (95) Streissguth AP, Barr HM, Sampson PD, Bookstein FL, Darby BL. Neurobehavioral effects of prenatal alcohol: Part I. Research strategy. Neurotoxicol Teratol 1989; 11(5): (96) Sampson PD, Streissguth AP, Barr HM, Bookstein FL. Neurobehavioral effects of prenatal alcohol: Part II. Partial least squares analysis. Neurotoxicol Teratol 1989; 11(5): (97) Streissguth AP, Bookstein FL, Sampson PD, Barr HM. Neurobehavioral effects of prenatal alcohol: Part III. PLS analyses of neuropsychologic tests. Neurotoxicol Teratol 1989; 11(5): (98) Streissguth AP, Sampson PD, Olson HC, Bookstein FL, Barr HM, Scott M et al. Maternal drinking during pregnancy: attention and short-term memory in 14- year-old offspring--a longitudinal prospective study. Alcohol Clin Exp Res 1994; 18(1): (99) O'Callaghan FV, O'Callaghan M, Najman JM, Williams GM, Bor W. Prenatal alcohol exposure and attention, learning and intellectual ability at 14 years: a prospective longitudinal study. Early Hum Dev 2007; 83(2): (100) Goldschmidt L, Richardson GA, Cornelius MD, Day NL. Prenatal marijuana and alcohol exposure and academic achievement at age 10. Neurotoxicol Teratol 2004; 26(4): (101) Ioffe S, Chernick V. Development of the EEG between 30 and 40 weeks gestation in normal and alcohol-exposed infants. Dev Med Child Neurol 1988; 30(6):

55 (102) Jacobson JL, Jacobson SW, Sokol RJ, Ager JW, Jr. Relation of maternal age and pattern of pregnancy drinking to functionally significant cognitive deficit in infancy. Alcohol Clin Exp Res 1998; 22(2): (103) Barr HM, Bookstein FL, O'Malley KD, Connor PD, Huggins JE, Streissguth AP. Binge drinking during pregnancy as a predictor of psychiatric disorders on the Structured Clinical Interview for DSM-IV in young adult offspring. Am J Psychiatry 2006; 163(6): (104) Olsen J. Effects of moderate alcohol consumption during pregnancy on child development at 18 and 42 months. Alcohol Clin Exp Res 1994; 18(5): (105) Kilburn TR. Average use of Alcohol and Binge Drinking in Pregnancy: Neuropsychological effects at age 5 - With particular reference to Information Processing Time [PhD thesis from the Faculty of Health Sciences, University of Aarhus; 2008]. (106) Underberg M. Prenatal alcohol exposure: effects on children's attention functions [PhD thesis from the Faculty of Health Sciences, University of Aarhus; 2008]. (107) Bell R, Lumley J. Alcohol consumption, cigarette smoking and fetal outcome in Victoria, Community Health Stud 1989; 13(4): (108) Sampson PD, Bookstein FL, Barr HM, Streissguth AP. Prenatal alcohol exposure, birthweight, and measures of child size from birth to age 14 years. Am J Public Health 1994; 84(9): (109) Tolo KA, Little RE. Occasional binges by moderate drinkers: implications for birth outcomes. Epidemiology 1993; 4(5): (110) Whitehead N, Lipscomb L. Patterns of alcohol use before and during pregnancy and the risk of small-for-gestational-age birth. Am J Epidemiol 2003; 158(7): (111) Pascoe JM, Kokotailo PK, Broekhuizen FF. Correlates of multigravida women's binge drinking during pregnancy. A longitudinal study. Arch Pediatr Adolesc Med 1995; 149(12): (112) O'Callaghan FV, O'Callaghan M, Najman JM, Williams GM, Bor W. Maternal alcohol consumption during pregnancy and physical outcomes up to 5 years of age: a longitudinal study. Early Hum Dev 2003; 71(2): (113) Godel JC, Pabst HF, Hodges PE, Johnson KE, Froese GJ, Joffres MR. Smoking and caffeine and alcohol intake during pregnancy in a northern population: effect on fetal growth. CMAJ 1992; 147(2):

56 (114) Jones KL, Chernoff GF, Kelly CD. Outcome of pregnancy in women who "binge" drink during the first trimester of pregnancy. [abstract] Clinical Research 1984; 32:114A. (115) Yuan W, Sorensen HT, Basso O, Olsen J. Prenatal maternal alcohol consumption and hospitalization with asthma in childhood: a population-based follow-up study. Alcohol Clin Exp Res 2004; 28(5): (116) Olsen J. Moderate alcohol consumption in pregnancy and subsequent lefthandedness. Scand J Soc Med 1995; 23(3): (117) Davis PJ, Partridge JW, Storrs CN. Alcohol consumption in pregnancy. How much is safe? Arch Dis Child 1982; 57(12): (118) Romitti PA, Sun L, Honein MA, Reefhuis J, Correa A, Rasmussen SA. Maternal periconceptional alcohol consumption and risk of orofacial clefts. Am J Epidemiol 2007; 166(7): (119) Grewal J, Carmichael SL, Ma C, Lammer EJ, Shaw GM. Maternal periconceptional smoking and alcohol consumption and risk for select congenital anomalies. Birth Defects Res A Clin Mol Teratol 2008; 82(7): (120) Williams LJ, Correa A, Rasmussen S. Maternal lifestyle factors and risk for ventricular septal defects. Birth Defects Res A Clin Mol Teratol 2004; 70(2): (121) Carmichael SL, Shaw GM, Yang W, Lammer EJ. Maternal periconceptional alcohol consumption and risk for conotruncal heart defects. Birth Defects Res A Clin Mol Teratol 2003; 67(10): (122) Adams MM, Mulinare J, Dooley K. Risk factors for conotruncal cardiac defects in Atlanta. J Am Coll Cardiol 1989; 14(2): (123) Plant ML, Plant MA. Maternal use of alcohol and other drugs during pregnancy and birth abnormalities: further results from a prospective study. Alcohol Alcohol 1988; 23(3): (124) Slickers JE, Olshan AF, Siega-Riz AM, Honein MA, Aylsworth AS. Maternal Body Mass Index and Lifestyle Exposures and the Risk of Bilateral Renal Agenesis or Hypoplasia: The National Birth Defects Prevention Study. Am J Epidemiol 2008; 168(11): (125) Iyasu S, Randall LL, Welty TK, Hsia J, Kinney HC, Mandell F et al. Risk factors for sudden infant death syndrome among northern plains Indians. JAMA 2002; 288(21):

57 (126) Olsen J, Melbye M, Olsen SF, Sorensen TI, Aaby P, Andersen AM et al. The Danish National Birth Cohort--its background, structure and aim. Scand J Public Health 2001; 29(4): (127) Greenland S, Pearl J, Robins JM. Causal diagrams for epidemiologic research. Epidemiology 1999; 10(1): (128) Glymour MM, Greenland S. Causal Diagrams. In: Rothman KJ, Greenland S, Lash TL, editors. Modern Epidemiology. 3rd ed. Philadelphia: Lippincott Williams & Wilkins; (129) Hernan MA, Hernandez-Diaz S, Werler MM, Mitchell AA. Causal knowledge as a prerequisite for confounding evaluation: an application to birth defects epidemiology. Am J Epidemiol 2002; 155(2): (130) Greenland S. Applications of Stratified Analysis Methods. In: Rothman KJ, Greenland S, Lash TL, editors. Modern Epidemiology. 3rd ed. Philadelphia: Lippincott Williams & Wilkins; (131) Jensen TK, Hjollund NH, Henriksen TB, Scheike T, Kolstad H, Giwercman A et al. Does moderate alcohol consumption affect fertility? Follow up study among couples planning first pregnancy. BMJ 1998; 317(7157): (132) Henriksen TB, Hjollund NH, Jensen TK, Bonde JP, Andersson AM, Kolstad H et al. Alcohol consumption at the time of conception and spontaneous abortion. Am J Epidemiol 2004; 160(7): (133) Littner Y, Bearer CF. Detection of alcohol consumption during pregnancy-- current and future biomarkers. Neurosci Biobehav Rev 2007; 31(2): (134) Midanik LT. Validity of self-reported alcohol use: a literature review and assessment. Br J Addict 1988; 83(9): (135) Kesmodel U, Olsen SF. Self reported alcohol intake in pregnancy: comparison between four methods. J Epidemiol Community Health 2001; 55(10): (136) Robles N, Day NL. Recall of alcohol consumption during pregnancy. J Stud Alcohol 1990; 51(5): (137) Alvik A, Haldorsen T, Lindemann R. Consistency of reported alcohol use by pregnant women: anonymous versus confidential questionnaires with item nonresponse differences. Alcohol Clin Exp Res 2005; 29(8): (138) Ekholm O. Influence of the recall period on self-reported alcohol intake. Eur J Clin Nutr 2004; 58(1):

58 (139) Olsen J, Tuntiseranee P. Is moderate alcohol intake in pregnancy associated with the craniofacial features related to the fetal alcohol syndrome? Scand J Soc Med 1995; 23(3): (140) Kesmodel U, Kesmodel PS, Larsen A, Secher NJ. Use of alcohol and illicit drugs among pregnant Danish women, Scand J Public Health 2003; 31(1):5-11. (141) Streissguth AP, Darby BL, Barr HM, Smith JR, Martin DC. Comparison of drinking and smoking patterns during pregnancy over a six-year interval. Am J Obstet Gynecol 1983; 145(6): (142) Serdula M, Williamson DF, Kendrick JS, Anda RF, Byers T. Trends in alcohol consumption by pregnant women through JAMA 1991; 265(7): (143) Floyd RL, Sidhu JS. Monitoring prenatal alcohol exposure. Am J Med Genet 2004; 127C(1):3-9. (144) Nohr EA, Frydenberg M, Henriksen TB, Olsen J. Does low participation in cohort studies induce bias? Epidemiology 2006; 17(4): (145) Kesmodel U. [Alcohol during pregnancy: consumption, attitudes, information and tracing]. Ugeskr Laeger 2003; 165(47): (146) Olsen J, Frische G, Kirchheiner H, Poulsen AO. [Alcohol-drinking habits among pregnant women in Odense ]. Ugeskr Laeger 1987; 149(21): (147) Ebrahim SH, Luman ET, Floyd RL, Murphy CC, Bennett EM, Boyle CA. Alcohol consumption by pregnant women in the United States during Obstet Gynecol 1998; 92(2): (148) Ebrahim SH, Diekman ST, Floyd RL, Decoufle P. Comparison of binge drinking among pregnant and nonpregnant women, United States, Am J Obstet Gynecol 1999; 180(1 Pt 1):1-7. (149) Naimi TS, Lipscomb LE, Brewer RD, Gilbert BC. Binge drinking in the preconception period and the risk of unintended pregnancy: implications for women and their children. Pediatrics 2003; 111(5 Part 2): (150) Gladstone J, Levy M, Nulman I, Koren G. Characteristics of pregnant women who engage in binge alcohol consumption. CMAJ 1997; 156(6): (151) Flynn HA, Marcus SM, Barry KL, Blow FC. Rates and correlates of alcohol use among pregnant women in obstetrics clinics. Alcohol Clin Exp Res 2003; 27(1):

59 (152) Than LC, Honein MA, Watkins ML, Yoon PW, Daniel KL, Correa A. Intent to become pregnant as a predictor of exposures during pregnancy: is there a relation? J Reprod Med 2005; 50(6): (153) Rasch V. Cigarette, alcohol, and caffeine consumption: risk factors for spontaneous abortion. Acta Obstet Gynecol Scand 2003; 82(2): (154) Abel EL. Maternal alcohol consumption and spontaneous abortion. Alcohol Alcohol 1997; 32(3): (155) Long MG, Waterson EJ, MacRae KD, Murray Lyon IM. Alcohol consumption and the risk of first trimester miscarriage. J Obstet Gynaecol 1994; 14: (156) Armstrong BG, McDonald AD, Sloan M. Cigarette, alcohol, and coffee consumption and spontaneous abortion. Am J Public Health 1992; 82(1): (157) Anokute CC. Epidemiology of spontaneous abortions: the effects of alcohol consumption and cigarette smoking. J Natl Med Assoc 1986; 78(8): (158) Kline J, Shrout P, Stein Z, Susser M, Warburton D. Drinking during pregnancy and spontaneous abortion. Lancet 1980; 2(8187): (159) Randall CL, Taylor J, Walker DW. Ethanol-induced malformations in mice. Alcohol Clin Exp Res 1977; 1(3): (160) Apgar BS, Churgay CA. Spontaneous abortion. Prim Care 1993; 20(3): (161) Kaufman MH. The teratogenic effects of alcohol following exposure during pregnancy, and its influence on the chromosome constitution of the preovulatory egg. Alcohol Alcohol 1997; 32(2): (162) Kaufman MH. Ethanol-induced chromosomal abnormalities at conception. Nature 1983; 302(5905): (163) Kaufman MH, Bain IM. The development potential of ethanol-induced monosomic and trisomic conceptuses in the mouse. J Exp Zool 1984; 231(1): (164) Kaufman MH. An hypothesis regarding the origin of aneuploidy in man: indirect evidence from an experimental model. J Med Genet 1985; 22(3): (165) Cavallo F, Russo R, Zotti C, Camerlengo A, Ruggenini AM. Moderate alcohol consumption and spontaneous abortion. Alcohol Alcohol 1995; 30(2): (166) Russell M, Skinner JB. Early measures of maternal alcohol misuse as predictors of adverse pregnancy outcomes. Alcohol Clin Exp Res 1988; 12(6):

60 (167) Faden VB, Graubard BI, Dufour M. The relationship of drinking and birth outcome in a US national sample of expectant mothers. Paediatr Perinat Epidemiol 1997; 11(2): (168) Kesmodel U, Wisborg K, Olsen SF, Henriksen TB, Secher NJ. Moderate alcohol intake during pregnancy and the risk of stillbirth and death in the first year of life. Am J Epidemiol 2002; 155(4): (169) Sokol RJ, Miller SI, Reed G. Alcohol abuse during pregnancy: an epidemiologic study. Alcohol Clin Exp Res 1980; 4(2): (170) Duncan JR, Randall LL, Belliveau RA, Trachtenberg FL, Randall B, Habbe D et al. The Effect of Maternal Smoking and Drinking During Pregnancy Upon (3)H- Nicotine Receptor Brainstem Binding in Infants Dying of the Sudden Infant Death Syndrome: Initial Observations in a High Risk Population. Brain Pathol 2008; 18(1): (171) Alm B, Wennergren G, Norvenius G, Skjaerven R, Oyen N, Helweg-Larsen K et al. Caffeine and alcohol as risk factors for sudden infant death syndrome. Nordic Epidemiological SIDS Study. Arch Dis Child 1999; 81(2): (172) Blair PS, Fleming PJ, Bensley D, Smith I, Bacon C, Taylor E et al. Smoking and the sudden infant death syndrome: results from case-control study for confidential inquiry into stillbirths and deaths in infancy. Confidential Enquiry into Stillbirths and Deaths Regional Coordinators and Researchers. BMJ 1996; 313(7051): (173) Kandall SR, Gaines J. Maternal substance use and subsequent sudden infant death syndrome (SIDS) in offspring. Neurotoxicol Teratol 1991; 13(2): (174) Ponnappa BC, Rubin E. Modeling alcohol's effects on organs in animal models. Alcohol Res Health 2000; 24(2): (175) Burd L, Deal E, Rios R, Adickes E, Wynne J, Klug MG. Congenital heart defects and fetal alcohol spectrum disorders. Congenit Heart Dis 2007; 2(4): (176) Streissguth AP, Clarren SK, Jones KL. Natural history of the fetal alcohol syndrome: a 10-year follow-up of eleven patients. Lancet 1985; 2(8446): (177) Loser H, Majewski F. Type and frequency of cardiac defects in embryofetal alcohol syndrome. Report of 16 cases. Br Heart J 1977; 39(12): (178) Smith DF, Sandor GG, MacLeod PM, Tredwell S, Wood B, Newman DE. Intrinsic defects in the fetal alcohol syndrome: studies on 76 cases from British Columbia and the Yukon Territory. Neurobehav Toxicol Teratol 1981; 3(2):

61 (179) Majewski F. Alcohol embryopathy: some facts and speculations about pathogenesis. Neurobehav Toxicol Teratol 1981; 3(2): (180) Halliday HL, Reid MM, McClure G. Results of heavy drinking in pregnancy. Br J Obstet Gynaecol 1982; 89(11): (181) Beattie JO, Day RE, Cockburn F, Garg RA. Alcohol and the fetus in the west of Scotland. Br Med J (Clin Res Ed ) 1983; 287(6384): (182) Webster WS, Germain MA, Lipson A, Walsh D. Alcohol and congenital heart defects: an experimental study in mice. Cardiovasc Res 1984; 18(6): (183) Beauchemin RR, Jr., Gartner LP, Provenza DV. Alcohol induced cardiac malformations in the rat. Anat Anz 1984; 155(1-5): (184) Adickes ED, Mollner TJ, Makoid MC. Ethanol-induced teratogenic alterations in developing cardiomyocytes in culture. Alcohol Alcohol Suppl 1993; 2: (185) Menegola E, Broccia ML, Di Renzo F, Giavini E. Acetaldehyde in vitro exposure and apoptosis: a possible mechanism of teratogenesis. Alcohol 2001; 23(1): (186) Pullarkat RK. Hypothesis: prenatal ethanol-induced birth defects and retinoic acid. Alcohol Clin Exp Res 1991; 15(3): (187) Shaw GM, Malcoe LH, Swan SH, Cummins SK, Schulman J. Congenital cardiac anomalies relative to selected maternal exposures and conditions during early pregnancy. Eur J Epidemiol 1992; 8(5): (188) Tikkanen J, Heinonen OP. Risk factors for cardiovascular malformations in Finland. Eur J Epidemiol 1990; 6(4): (189) Cedergren MI, Selbing AJ, Kallen BA. Risk factors for cardiovascular malformation--a study based on prospectively collected data. Scand J Work Environ Health 2002; 28(1): (190) McDonald AD, Armstrong BG, Sloan M. Cigarette, alcohol, and coffee consumption and congenital defects. Am J Public Health 1992; 82(1): (191) Ferencz C, Loffredo CA, Correa-Villasenor A, Wilson PD. Genetic and environmental risk factors of major cardiovascular malformations: the Baltimore- Washington Infant Study Armonk, NY: Futura Publishing Company, Inc; (192) Damgaard I.N., Jensen T.K., the Nordic Cryptorchidism Study Group, Petersen J.H., Skakkebæk N.E., Toppari J et al. Cryptorchidism and Maternal Alcohol Consumption during Pregnancy. Environmental Health Perspectives. In press

62 (193) Carbone P, Giordano F, Nori F, Mantovani A, Taruscio D, Lauria L et al. The possible role of endocrine disrupting chemicals in the aetiology of cryptorchidism and hypospadias: a population-based case-control study in rural Sicily. Int J Androl 2007; 30(1):3-13. (194) McBride ML, Van den SN, Lamb CW, Gallagher RP. Maternal and gestational factors in cryptorchidism. Int J Epidemiol 1991; 20(4): (195) Kurahashi N, Kasai S, Shibata T, Kakizaki H, Nonomura K, Sata F et al. Parental and neonatal risk factors for cryptorchidism. Med Sci Monit 2005; 11(6):CR274- CR283. (196) Davies TW, Williams DR, Whitaker RH. Risk factors for undescended testis. Int J Epidemiol 1986; 15(2): (197) Biggs ML, Baer A, Critchlow CW. Maternal, delivery, and perinatal characteristics associated with cryptorchidism: a population-based case-control study among births in Washington State. Epidemiology 2002; 13(2): (198) Berkowitz GS, Lapinski RH. Risk factors for cryptorchidism: a nested casecontrol study. Paediatr Perinat Epidemiol 1996; 10(1): (199) Mori M, Davies TW, Tsukamoto T, Kumamoto Y, Fukuda K. Maternal and other factors of cryptorchidism--a case-control study in Japan. Kurume Med J 1992; 39(2): (200) Jensen MS, Bonde JP, Olsen J. Prenatal alcohol exposure and cryptorchidism. Acta Paediatr 2007; 96(11): (201) Moller H, Skakkebaek NE. Testicular cancer and cryptorchidism in relation to prenatal factors: case-control studies in Denmark. Cancer Causes Control 1997; 8(6): (202) Mongraw-Chaffin ML, Cohn BA, Cohen RD, Christianson RE. Maternal smoking, alcohol consumption, and caffeine consumption during pregnancy in relation to a son's risk of persistent cryptorchidism: a prospective study in the Child Health and Development Studies cohort, Am J Epidemiol 2008; 167(3): (203) Nagata C, Iwasa S, Shiraki M, Sahashi Y, Shimizu H. Association of maternal fat and alcohol intake with maternal and umbilical hormone levels and birth weight. Cancer Sci 2007; 98(6): (204) Petridou E, Katsouyanni K, Spanos E, Skalkidis Y, Panagiotopoulou K, Trichopoulos D. Pregnancy estrogens in relation to coffee and alcohol intake. Ann Epidemiol 1992; 2(3):

63 (205) Wuu J, Hellerstein S, Lipworth L, Wide L, Xu B, Yu GP et al. Correlates of pregnancy oestrogen, progesterone and sex hormone-binding globulin in the USA and China. Eur J Cancer Prev 2002; 11(3): (206) Stevens RG, Cohen RD, Terry MB, Lasley BL, Siiteri P, Cohn BA. Alcohol consumption and serum hormone levels during pregnancy. Alcohol 2005; 36(1): (207) Tikkanen J, Heinonen OP. Risk factors for atrial septal defect. Eur J Epidemiol 1992; 8(4): (208) Tikkanen J, Heinonen OP. Risk factors for ventricular septal defect in Finland. Public Health 1991; 105(2): (209) Tikkanen J, Heinonen OP. Maternal exposure to chemical and physical factors during pregnancy and cardiovascular malformations in the offspring. Teratology 1991; 43(6): (210) Makela P, Fonager K, Hibell B, Nordlund S, Sabroe S, Simpura J. Episodic heavy drinking in four Nordic countries: a comparative survey. Addiction 2001; 96(11): (211) Hernan MA, Hernandez-Diaz S, Robins JM. A structural approach to selection bias. Epidemiology 2004; 15(5): (212) Rothman K, Greenland S, Lash TL. Validty in Epidemiologic Studies. Modern Epidemiology. 3rd ed. Philadelphia: Lippincott Williams & Wilkins; (213) Olsen J. Options in making use of pregnancy history in planning and analysing studies of reproductive failure. J Epidemiol Community Health 1994; 48(2): (214) Rothman K. Biases in Study Design. Epidemiology - An introduction. Oxford University Press; (215) Dawson DA. Volume of ethanol consumption: effects of different approaches to measurement. J Stud Alcohol 1998; 59(2): (216) Rehm J. Measuring quantity, frequency, and volume of drinking. Alcohol Clin Exp Res 1998; 22(2 Suppl):4S-14S. (217) McCann SE, Marshall JR, Trevisan M, Russell M, Muti P, Markovic N et al. Recent alcohol intake as estimated by the Health Habits and History Questionnaire, the Harvard Semiquantitative Food Frequency Questionnaire, and a more detailed alcohol intake questionnaire. Am J Epidemiol 1999; 150(4):

64 (218) Greenland S, Morgenstern H. Confounding in health research. Annu Rev Public Health 2001; 22: (219) Balle J, Olofsson MJ, Hilden J. [Cannabis and pregnancy]. Ugeskr Laeger 1999; 161(36): (220) Strobino BR, Kline J, Stein Z. Chemical and physical exposures of parents: effects on human reproduction and offspring. Early Hum Dev 1978; 1(4): (221) Morch LS, Becker U, Olsen J, Tjonneland AM, Gronbaek MN. [Should the sensible drinking limits for adults be changed?]. Ugeskr Laeger 2005; 167(40): (222) Grandjean P, Bailar JC, Gee D, Needleman HL, Ozonoff DM, Richter E et al. Implications of the Precautionary Principle in research and policy-making. Am J Ind Med 2004; 45(4): (223) Tickner JA, Kriebel D, Wright S. A compass for health: rethinking precaution and its role in science and public health. Int J Epidemiol 2003; 32(4): (224) Tolstrup J, Jensen MK, Tjonneland A, Overvad K, Mukamal KJ, Gronbaek M. Prospective study of alcohol drinking patterns and coronary heart disease in women and men. BMJ 2006; 332(7552): (225) Tolstrup JS, Jensen MK, Tjonneland A, Overvad K, Gronbaek M. Drinking pattern and mortality in middle-aged men and women. Addiction 2004; 99(3):

65 P A P E R I Strandberg-Larsen K, Nybo Andersen AM, Olsen J, Nielsen NR, and Grønbæk M: Do women give the same information on binge drinking during pregnancy when asked repeatedly? European Journal of Clinical Nutrition 2006; 60:

66 European Journal of Clinical Nutrition (2006) 60, & 2006 Nature Publishing Group All rights reserved /06 $ ORIGINAL ARTICLE Do women give the same information on binge drinking during pregnancy when asked repeatedly? K Strandberg-Larsen 1, A-MN Andersen 1, J Olsen 2, NR Nielsen 1,2 and M Grønbæk 1 1 National Institute of Public Health, Copenhagen, Denmark and 2 Department of Epidemiology, UCLA School of Public Health, Los Angeles, CA, USA Objective: To study if pregnant women give the same answers to questions on frequency and timing of binge drinking when asked more than once during and after pregnancy. Design: Cohort study. Setting: The Danish National Birth Cohort. Subjects: The study is based on pregnant women with repeated information on binge drinking during the early part of pregnancy and 8933 pregnant women with information on binge drinking during pregnancy weeks 30 36, obtained while pregnant and 6 months after delivery. Results: More women reported binge drinking, if the interview took place close to the period in question. As the report of binge drinking was highest in the first of two interviews referring to the same period, as well as women who participated in the first interview in pregnancy week 12 or earlier reported more binge drinking compared to women who participated in the interview later in pregnancy. Conclusions: Self-reported information on binge drinking is more frequently under-reported when the recall period is long. To improve the validity of data on binge drinking, future birth cohorts should obtain information several times during pregnancy. Sponsorship: The Danish National Board of Health and the Health Insurance Foundation. European Journal of Clinical Nutrition (2006) 60, doi: /sj.ejcn ; published online 24 May 2006 Keywords: alcohol drinking; bias; interviews; pregnancy; recall Introduction Animal models have indicated that sporadic high intake of alcohol (binge drinking) is more harmful to the developing fetus than if the same amount of alcohol is spread out over several days (Pierce and West, 1986; Bonthius et al., 1988; Correspondence: K Strandberg-Larsen, Centre for Alcohol Research, National Institute of Public Health, Øster Farimagsgade 5 A, 2nd floor, DK-1399 Copenhagen K, Denmark. kal@niph.dk Guarantor: K Strandberg-Larsen. Contributors: KS-L contributed to the conception and design of the study, analysis and interpretation of data, and drafting the paper. A-MNA, JO, NRN, and MG contributed to the conception and design of the study and to critically revising the manuscript. A-MNA and JO contributed to the design of the Danish National Birth Cohort and acquisition of data. A-MNA, JO and MG have designed the questions on number and timing of binge drinking used in the present study. All contributors have approved the final version of the manuscript. Received 25 August 2005; revised 17 December 2005; accepted 15 February 2006; published online 24 May 2006 West et al., 1990; Goodlett and Eilers, 1997). Almost none of the existing studies of the association between binge drinking and adverse pregnancy outcomes have included information on number and timing of binge episodes even though the effects of binge drinking are hypothesized to be time-specific (Allebeck and Olsen, 1998). Leaving out timing of binge drinking may lead to serious underestimates of the effects of binge drinking in periods of gestation where the fetus is, owing to its developmental status, especially vulnerable to harmful effects of binge drinking. If such periods exist, it may explain why previous studies have failed to find any association between binge drinking and birth weight, length at birth, head circumference, gestational age, Apgar score and malformations. (Tolo and Little, 1993; Olsen and Tuntiseranee, 1995; Passaro et al., 1996; Kesmodel, 2001; Whitehead and Lipscomb, 2003). Valid data on when in pregnancy binge drinking occurs are needed, if we want to clarify the potential time-specific health hazards of binge drinking. The Danish National Birth Cohort of approximately pregnant women and their

67 pregnancy outcomes covers information on number and point(s) in pregnancy of binge drinking. These data constitute an excellent opportunity to study the potential deleterious effects of binge drinking and time-specific hazards, provided that the women gave valid information. In this paper, we assess the validity of the collected data by examining if women give the same information on binge drinking during pregnancy when asked repeatedly, and if the length of the recall period influences the self-reported occurrence of binge drinking. Materials and methods Study population The study used data from the Danish National Birth Cohort (DNBC). Enrolment into the cohort was organized through the general practitioners, and from 1996 to 2002, approximately pregnant women and their outcomes of pregnancy were recruited to the cohort. Women provided information on exposures during pregnancy by means of three computer-assisted telephone interviews, scheduled to take place in pregnancy weeks 12 and 30 (first and second interview), and 6 months after delivery (third interview). The interviews were given up if the pregnancy had ended before the scheduled interviews or if no contact was established within four attempts. Details on the DNBC regarding study design, recruitment and procedures have been published elsewhere (Olsen et al., 2001; Nybo Andersen and Olsen, 2002). The present study was based on two subgroups of women. The first consisted of the women who gave information on binge drinking during the early part of pregnancy in the first and again in the second interview. The second subgroup consisted of the 8933 women who, in the second and third interview, gave information on binge drinking during pregnancy weeks Information on binge drinking The questions on number and points in pregnancy of binge drinking, including the very first part of pregnancy, were identical in the first and second interview. Binge drinking was defined as an alcohol intake of five or more drinks on one occasion or on an evening. Points in pregnancy of binge drinking were reported in commenced gestational weeks (pregnancy weeks). The question in the third interview was the same, but asked about binge drinking in the period from pregnancy week 30 until delivery. Women were categorized as binge drinkers if they reported at least one episode of binge drinking. Number of binge episodes reported in the first and second interview was categorized as 0, 1, 2 and 3 þ episodes. Repeatability of self-reported binge drinking K Strandberg-Larsen et al Statistical analysis The agreement was assessed by four methods: proportion of agreement, kappa value and two separate indices for positive and negative proportion of agreement. The denominator of the separate index for positive proportion of agreement was the average of the positive responses in the two succeeding interviews, as the denominator of the negative proportion of agreement was similar to the average of the negative responses. Furthermore, a measure of bias in disagreement between two succeeding interviews was calculated as ((Yes Interview 1,No Interview 2 ) (Yes Interview 2,No Interview 1 ))/N (Kesmodel and Frydenberg, 2004). The influence of the length of the recall period on the reporting of binge drinking was examined by comparing the proportion of binge drinkers among women who answered the first interview early (in pregnancy week 12 or earlier) to the proportion of binge drinkers among those who were interviewed later (after pregnancy week 12). We further stratified the analyses of agreement between the first and second interview according to the time gap between answering the interviews (10 weeks or less versus more than 10 weeks). The week-byweek analyses were stratified according to changed or unchanged gestational age, to assess if changes in the estimation of gestational age between the first and second interview explained the potential disagreement between the reported points in pregnancy of binge drinking. The disagreement between the number of binge episodes reported in the first and second interview was analysed and described by the rank-invariant method (Svensson, 1997, 1998). Systematic inter-interview differences were divided into relative position (RP) and relative concentration (RC), which display the disagreement attributable to a consequent underestimation of number of episodes in one of the interviews relative to the other and the disagreement attributable to if the classification in one of the interviews is concentrated to a limited part of the number of episodes relative to the other. Possible values of RP and RC range from 1 to 1 and values close to zero indicate negligible bias between the interviews. The standard errors (s.e.) of the RP and RC were estimated by the jackknife technique (Svensson, 1998). Results The agreement of information on binge drinking given in the first and second interview was 0.85, with a kappa value of 0.56 and a positive agreement of 0.65 (Table 1). More women reported binge drinking in the first interview compared to in the second interview, the estimate of bias ¼ 5.9%. Very few women reported binge drinking in the period from pregnancy weeks 30 36; in the second interview, 84 women reported binge drinking compared to only 37 women in the third interview (Table 1) European Journal of Clinical Nutrition

68 1296 Table 1 Agreement of self-reported binge drinking during the early part of pregnancy and pregnancy weeks Binge drinking in the first interview Binge drinking in the second interview Yes No Total Yes No Total Proportion of agreement 0.85 Kappa value 0.56 Positive proportion of agreement 0.65 Negative proportion of agreement 0.91 Bias 5.9 Binge drinking in the second interview Binge drinking in the third interview Yes No Total Yes No Total Proportion of agreement 0.99 Kappa value 0.11 Positive proportion of agreement 0.12 Negative proportion of agreement 0.99 Bias 0.5 Proportion of women reporting binge drinking (%) Repeatability of self-reported binge drinking K Strandberg-Larsen et al First interview Second interview Weeks of pregnancy Figure 1 Proportion of women who reported binge drinking during the first 20 weeks of pregnancy, among women for whom information was available in the two interviews during pregnancy. Reports on points in pregnancy of binge drinking A higher proportion of women reported at least one episode of binge drinking during the early weeks of pregnancy in the first interview compared to in the second interview (Figure 1). The agreement of the reported timing of binge episodes in the first and second interview was assessed by the week-by-week agreement. The proportion of agreement ranged from 0.90 to 1.00, whereas the kappa values ranged from 0.10 to 0.54 (data not shown). Changes in the estimation of the gestational age in the period between Table 2 The joint frequency distribution of the number of binge drinking episodes reported in the two interviews during pregnancy Number of binge episodes in the first interview Number of binge episodes in the second interview þ Total 3 þ Total Proportion of women reporting binge drinking (%) Answered the interview in week 12 or earlier Answered the interview in week 13 or later Weeks of pregnancy Figure 2 Proportion of binge drinkers in the first interview among women interviewed in pregnancy week 12 or earlier and women interviewed later than pregnancy week 12. answering the first and second interview had no influence on the week-by-week agreement (data not shown). Reports on number of binge drinking episodes The joint frequency distribution of the reported number of binge episodes obtained in the first and second interview is shown in Table 2. Eighty-one per cent of the women reported the same number of binge drinking episodes in the first and second interview. The marginal distributions between the first and second interview differed, which implies systematic difference between the two interviews, which were attributable to a slight underestimation of binge drinking in the second interview compared to the first interview (RP ¼ 0.058, s.e. ¼ 0.001). The reported number of binge drinking episodes in one of the interviews was not concentrated to a limited part of the reported number of binge episodes in the other interview (RC ¼ 0.007, s.e. ¼ 0.001). Importance of the length of the recall period Women who answered the first interview in pregnancy week 12 or earlier reported more often binge drinking during the first 4 weeks of pregnancy compared to women who were interviewed later during pregnancy (Figure 2). European Journal of Clinical Nutrition

69 The agreement of the information on binge drinking given in the first and second interview was also influenced by the time gap between the two interviews. The agreement was higher among women with a time gap of 10 weeks or less compared with women where the time gap was greater than 10 weeks, kappa ¼ 0.63 versus kappa ¼ 0.54 (Po0.0001) (data not shown). Discussion These results show that pregnant women recall binge drinking better if the data collection takes place close in time to the reported binge episodes. In our study, the pregnant women reported differently on number and timing of binge drinking episodes in two interviews placed in midand late pregnancy. The lack of repeatability of self-reported information on binge drinking obtained in two subsequent interviews is presumably attributable to the variant recall periods in the interviews. The repeatability of self-reported information on pregnancy-related binge drinking obtained by interview twice during pregnancy has not been studied in detail before. One study examined the week-by-week agreement between information on binge drinking during pregnancy obtained by questionnaire and a subsequent face-to-face interview (Kesmodel and Frydenberg, 2004). In contrast to our results, the variation in time between filling in the questionnaire and answering the interview had no influence on the agreement between these two methods of obtaining information on binge drinking (Kesmodel and Frydenberg, 2004). In line with the results of our study, the week-by-week agreement between the answers in the questionnaire and interview was low. Our results regarding the importance of the recall period are supported by a study based on the Danish Health Interview Survey, which showed that the selfreported average intake of alcohol systematically decreased as the recall period increased among non-pregnant respondents (Ekholm, 2004). Collecting information on binge drinking by telephone interviews is a strength of the present study because in comparison to self-administered questionnaires, interviews in general result in a higher participation rate and a higher response rate to the specific questions regarding binge drinking (Kesmodel, 2001; Kesmodel and Frydenberg, 2004). It is a strength of the present study that the kappa value is accompanied by separate individual values of positive and negative proportion of agreement as the correction factor in the kappa index adjusts the results for the discrepancies in the positive and negative agreement, which in this study are large and the cause to the fact that the high proportions of agreements are followed by much lower kappa values (Cicchetti and Feinstein, 1990). None of the interviews are necessarily valid measures of the actual occurrence of binge drinking. For self-reported information on average alcohol intake, it is generally Repeatability of self-reported binge drinking K Strandberg-Larsen et al assumed that the highest reported intake is the most valid measure of the actual intake as few will report an alcohol intake they did not have (Kesmodel and Olsen, 2001). If this assumption also applies to binge drinking during pregnancy, it implies that the answers given in the first interview are the most valid measure of the actual occurrence of binge drinking during the early part of pregnancy. Similarly, the second interview is presumably the most valid measure of binge drinking during pregnancy weeks However, the reported number of binge episodes in the first and second interview may be an underestimation of the actual number of binge drinking episodes, especially among women with a high actual occurrence of binge drinking. In the interpretation of the existing epidemiological studies regarding the harmful effects of binge drinking, our conclusion regarding the decline in the accuracy of the reported information on binge drinking as the recall period increases is important. The major part of the existing studies has collected information on binge drinking in the second half of pregnancy or subsequent to birth (Tolo and Little, 1993; Olsen, 1994; Pascoe et al., 1995; Passaro et al., 1996; Iyasu et al., 2002; Whitehead and Lipscomb, 2003) and therefore the information on binge drinking is most likely underestimated. Due to this underestimation, actual exposed women may be categorized as non-exposed and the estimates could probably be biased towards the null-value. This may explain why the existing studies have shown little or no detrimental effects of binge drinking during pregnancy. The severity of underreporting in a follow-up design depends upon the frequency of binge drinking. If this frequency is low, the number of binge drinkers who are miscategorized as not-exposed will be low and will be a small fraction of the not-exposed. Thus, the likelihood of detecting an adverse effect of binge drinking is decreased, but the bias is limited. The lack of ability of women to report the same timing of binge drinking when asked repeatedly is problematic, as it implies that the quality of the data on timing may be low and therefore harm future efforts to identify vulnerable time periods. To minimize underestimation of binge drinking in future cohort studies, special efforts should be made to obtain information on binge drinking as early as possible in pregnancy and use longitudinal measures to collect information on binge drinking during the remaining period of pregnancy while the women are still pregnant. Such an approach is not without problems, as the more we ask respondents to do, the more difficult it may be to recruit pregnant women to the study, and the more we may influence their way of living. Hence, a very thorough data collection may also in itself influence their drinking behaviour, which could not only limit the number of informative observations, but also cause confounding if the change in health behaviours is restricted to specific groups of women, and this type of confounding may be difficult to adjust for. More accurate data on timing of binge drinking may be obtained by asking for calendar time rather than 1297 European Journal of Clinical Nutrition

70 1298 pregnancy time, as binge drinking is related to specific occasions that are not recalled in pregnancy time. Acknowledgements This particular study is funded by grants from the Danish National Board of Health and the Health Insurance Foundation. The Danish National Research Foundation has established the Danish Epidemiology Science Centre that initiated and created the Danish National Birth Cohort. The cohort is furthermore a result of a major grant from this foundation. Additional support for the Danish National Birth Cohort is obtained from the Pharmacy Foundation, the Egmont Foundation, the March of Dimes Birth Defects Foundation and the Augustinus Foundation. References Repeatability of self-reported binge drinking K Strandberg-Larsen et al Allebeck P, Olsen J (1998). Alcohol and fetal damage. Alcohol Clin Exp Res 22, Bonthius DJ, Goodlett CR, West JR (1988). Blood alcohol concentration and severity of microencephaly in neonatal rats depend on the pattern of alcohol administration. Alcohol 5, Cicchetti DV, Feinstein AR (1990). High agreement but low kappa: II. Resolving the paradoxes. J Clin Epidemiol 43, Ekholm O (2004). Influence of the recall period on self-reported alcohol intake. Eur J Clin Nutr 58, Goodlett CR, Eilers AT (1997). Alcohol-induced Purkinje cell loss with a single binge exposure in neonatal rats: a stereological study of temporal windows of vulnerability. Alcohol Clin Exp Res 21, Iyasu S, Randall LL, Welty TK, Hsia J, Kinney HC, Mandell F et al. (2002). Risk factors for sudden infant death syndrome among northern plains Indians. JAMA 288, Kesmodel U (2001). Binge drinking in pregnancy frequency and methodology. Am J Epidemiol 154, Kesmodel U, Frydenberg M (2004). Binge drinking during pregnancy is it possible to obtain valid information on a weekly basis? Am J Epidemiol 159, Kesmodel U, Olsen SF (2001). Self reported alcohol intake in pregnancy: comparison between four methods. J Epidemiol Community Health 55, Nybo Andersen AM, Olsen J (2002). Do interviewers health beliefs and habits modify responses to sensitive questions? A study using data collected from pregnant women by means of computerassisted telephone interviews. Am J Epidemiol 155, Olsen J (1994). Effects of moderate alcohol consumption during pregnancy on child development at 18 and 42 months. Alcohol Clin Exp Res 18, Olsen J, Melbye M, Olsen SF, Sorensen TI, Aaby P, Andersen AM et al. (2001). The Danish National Birth Cohort its background, structure and aim. Scand J Public Health 29, Olsen J, Tuntiseranee P (1995). Is moderate alcohol intake in pregnancy associated with the craniofacial features related to the fetal alcohol syndrome? Scand J Soc Med 23, Pascoe JM, Kokotailo PK, Broekhuizen FF (1995). Correlates of multigravida women s binge drinking during pregnancy. A longitudinal study. Arch Pediatr Adolesc Med 149, Passaro KT, Little RE, Savitz DA, Noss J (1996). The effect of maternal drinking before conception and in early pregnancy on infant birthweight. The ALSPAC Study Team. Avon Longitudinal Study of Pregnancy and Childhood. Epidemiology 7, Pierce DR, West JR (1986). Blood alcohol concentration: a critical factor for producing fetal alcohol effects. Alcohol 3, Svensson E (1997). A coefficient of agreement adjusted for bias in paired ordered categorical data. Biometrical J 39, Svensson E (1998). Apllication of a rank-invariant method to evaluate reliability of ordered categorical assesments. J Epidemiol Biostat 3, Tolo KA, Little RE (1993). Occasional binges by moderate drinkers: implications for birth outcomes. Epidemiology 4, West JR, Goodlett CR, Bonthius DJ, Hamre KM, Marcussen BL (1990). Cell population depletion associated with fetal alcohol brain damage: mechanisms of BAC-dependent cell loss. Alcohol Clin Exp Res 14, Whitehead N, Lipscomb L (2003). Patterns of alcohol use before and during pregnancy and the risk of small-for-gestational-age birth. Am J Epidemiol 158, European Journal of Clinical Nutrition

71 P A P E R I I Strandberg-Larsen K, Nielsen NR, Nybo Andersen AM, Olsen J, and Grønbæk M: Characteristics of women who binge drink before and after they become aware of their pregnancy. European Journal of Epidemiology 2008; 23:

72 Eur J Epidemiol (2008) 23: DOI /s z PERINATAL EPIDEMIOLOGY Characteristics of women who binge drink before and after they become aware of their pregnancy Katrine Strandberg-Larsen Æ Naja Rod Nielsen Æ Anne-Marie Nybo Andersen Æ Jørn Olsen Æ Morten Grønbæk Received: 14 December 2007 / Accepted: 19 May 2008 / Published online: 14 June 2008 Ó Springer Science+Business Media B.V Abstract Background Consumption of high doses of alcohol on a single occasion (binge drinking) may harm the developing foetus and pregnant women are advised to avoid binge drinking while pregnant. We present characteristics of Danish women who binge drank in the pre-and post recognised part of their pregnancy. Methods During the years approximately 100,000 pregnant women were enrolled into the Danish National Birth Cohort. Women with information on binge drinking, time of recognition of pregnancy, age, reproductive history, marital status, smoking, occupational status, pre-pregnancy BMI, alcohol consumption before pregnancy, and mental disorders (n = 85,334) were included in the analyses. Results Approximately one quarter of the women reported binge drinking at least once during pregnancy; most of these in the pre-recognised part of pregnancy. Weekly alcohol consumption before pregnancy, single status and smoking were predictors for binge drinking in both the unrecognised and recognised part of pregnancy. Moreover, binge drinking in the pre-recognised part of K. Strandberg-Larsen (&) M. Grønbæk Centre for Alcohol Research, National Institute of Public Health, University of Southern Denmark, Øster Farimagsgade 5A, 2. Floor, DK-1399 Copenhagen K, Denmark kal@niph.dk N. Rod Nielsen National Institute of Public Health, University of Southern Denmark, Øster Farimagsgade 5A, 2. Floor, DK-1399 Copenhagen K, Denmark A.-M. Nybo Andersen Division of Epidemiology, University of Southern Denmark, J.B. Winsløws Vej 9B, DK-5000 Odense C, Denmark J. Olsen Division of Epidemiology, UCLA School of Public Health, CHS, Box , Los Angeles, CA , USA pregnancy was more common among women aged years, who were nulliparous, well educated in good jobs or skilled workers. Binge drinking after recognition of pregnancy was more common among women who were unintended pregnant, multiparous unskilled workers, had been unemployed for more than one year, or had mental/ neurotic disorder. Conclusions In order to prevent binge drinking during pregnancy, health care providers should target their efforts towards pregnant women as well as pregnancy-planners. It is important to be aware that women who binge drink before versus after the pregnancy is recognised have different social characteristics. Keywords Prevention Introduction Alcohol Binge drinking Pregnancy In many westernised countries, consumption of five or more drinks of alcohol on one occasion binge drinking is frequent among women at childbearing age [1 4]. However, animal models as well as epidemiological studies have suggested that intermittent exposure to high levels of alcohol, as would be expected with maternal binge drinking, may harm the offspring and cause important neurological damages [5 8]. Despite this knowledge and official warnings against binge drinking in pregnancy, binge drinking among pregnant women is increasing in the US and around 50% of pregnant Danes have at least one episode of binge drinking [9 14]. In order to develop targeted prevention strategies we need to know which women are prone to engage in this behaviour while pregnant, but we do not. The aim of this study is to identify lifestyles, socio-demographic factors, and aspects of reproductive history that are associated with binge 123

73 566 K. Strandberg-Larsen et al. drinking in the pre- and post-recognised part of pregnancy among Danish women. Methods Study population The Danish National Birth Cohort (DNBC) is a populationbased cohort of pregnant women and their children [15]. During pregnant women were approached at the first antenatal visit by the general practitioners who handed out information about the cohort as well as an informed consent form. To be eligible to participate in the DNBC the women had to be pregnant, intend to carry their pregnancy to term, have a permanent address in Denmark, and speak Danish well enough to participate in telephone interviews. The women were registered as participants when the study centre received the signed informed consent form. Participants provided information on exposures during the early part of pregnancy by means of a computer-assisted telephone interview, scheduled to occur around 12th 16th weeks of gestation. An English translation of the interview guide is available at For this study, we used data on the 90,165 women who participated in the interview. We excluded participants with no information on binge drinking (n = 343), timing of at least one of the binge episode (n = 836), time of pregnancy recognition (n = 403), parity (n = 81), marital status (n = 33), time to pregnancy (n = 287), smoking (n = 221), occupational status (n = 833), pre-pregnancy BMI (n = 1,391), mental disorders or neurosis (n = 65), or alcohol consumption prior to pregnancy (n = 338). Thus, the remaining 85,334 participants were included in the analyses. Information on binge drinking The women were asked: Think of your entire pregnancy including the very first period of pregnancy and tell me how many times have you consumed five or more drinks on one occasion?. If the women reported any binge drinking they were asked to state the week in which each binge drinking episode occurred. The interviewers were instructed to clarify that the pregnancy was calculated from the first day of the last menstrual period and when reporting on time in pregnancy this estimation of gestational age should be used. Besides giving information on timing of each binge drinking episodes the women also reported in what week the pregnancy was recognised. Information on timing of binge drinking and pregnancy recognition was combined to define whether binge drinking had occurred (yes or no) in the pre-recognised and recognised part of pregnancy. These categories are not mutually exclusive. The number of binge episodes in the recognised part of pregnancy was categorised as 1, 2 3, 4+ binge episodes. Analytic methods We used logistic regression models to estimate to what degree socio-demographic factors, reproductive history, and lifestyle factors are associated with binge drinking prior to and subsequent to pregnancy recognition. Age at conception, prior reproductive history, time to pregnancy, marital status, smoking during pregnancy, occupational status, weight status before pregnancy, alcohol consumption before pregnancy, week of pregnancy recognition, and self-reports on mental/neurotic disorders were categorised as in Table 1. All characteristics were included in univariate models as well as in multiple regressions. Separate analyses were made for binge drinking in the pre- and post-recognised part of pregnancy. The analyses regarding binge drinking in the prerecognised part of pregnancy was stratified according to whether or not the pregnancy was planned. This stratification was made to assess if differences in motivations to change lifestyle, and thereby attenuate alcohol behaviour may affect characteristics of women who binge drink prior to pregnancy recognition among planned vs. unplanned pregnancy. Finally, all analyses were stratified according to parity to assess if the associations differed for nulliparous and multiparous women. Results A total of 20,557 (24.1%) women reported binge drinking during pregnancy and 1,216 (5.8%) of these women reported four or more episodes of binge drinking. The majority of binge drinking took place in the pre-recognised part of pregnancy, since 18,436 women reported binge drinking prior to pregnancy recognition, while only 2,999 women reported binge drinking subsequent to their recognition of pregnancy. Characteristics of women who binge drink in the pre-recognised part of pregnancy Binge drinking in the pre-recognised part of pregnancy were more common in unplanned pregnancies (27.6%) than in the 88.8% planned pregnancies (20.9%). Binge drinking in the pre-recognised part of pregnancy correlates with women s age, prior reproductive history, time to pregnancy, marital status, smoking habits, occupational status, women s pre-pregnancy weight status, weekly average alcohol consumption before pregnancy, and week of pregnancy recognition, but not with presence of mental disorders or neurosis (Table 1). Women aged years 123

74 Characteristics of pregnant binge drinkers 567 Table 1 Characteristics of women who binge drink in the pre-recognized part of pregnancy, according to planning-status of pregnancy Binge drinking in the pre-recognized part of pregnancy Pregnancy-planners Non-planners N % Binge drinkers OR N % Binge (95% CI) a drinkers OR (95% CI) a Total population 75, , Age in years \ ( ) ( ) , ( ) 1, ( ) , (Reference) 2, (Reference) , ( ) 2, ( ) C35 8, ( ) 1, ( ) Prior reproductive history First-time pregnant women 26, ( ) 3, ( ) Nulliparous, history of induced abortion 3, ( ) ( ) Nulliparous, history of spontaneous abortion 5, ( ) ( ) Multiparous 39, (Reference) 5, (Reference) Time to pregnancy (months) B2 37, (Reference) , ( ) , ( ) [12 11, ( ) Marital status Married/cohabiting 75, (Reference) 8, (Reference) Single ( ) 1, ( ) Smoking during pregnancy Smoker 17, ( ) 3, ( ) Non-smoker 57, (Reference) 5, (Reference) Occupational status Higher grade professionals 7, ( ) ( ) Lower grade professionals 21, (Reference) 1, (Reference) Skilled workers 14, ( ) 1, ( ) Unskilled workers 18, ( ) 2, ( ) Students 9, ( ) 1, ( ) Unemployed [ 1 year 3, ( ) ( ) Weight status before pregnancy (BMI) Underweight (\18,5) 3, ( ) ( ) Normal (18,5 24) 51, (Reference) 6, (Reference) Overweight (25 29) 14, ( ) 1, ( ) Obese (C30) 6, ( ) ( ) Weekly alcohol consumption prior to pregnancy in drinks 0 9, (Reference) 1, (Reference) 1 24, ( ) 2, ( ) , ( ) 1, ( ) 4+ 22, ( ) 3, ( ) Week of recognition of pregnancy \4 4, ( ) ( ) , (Reference) 4, (Reference) , ( ) 2, ( ) C8 5, ( ) 1, ( ) a Odds ratios from a multivariate model including all the characteristics in the table 123

75 568 K. Strandberg-Larsen et al. old binge drank more often in the pre-recognised part of pregnancy than younger or older women. Among planned pregnancies the association with age differed for nulliparous and multiparous women, since age less than 25 decreased the odds of binge drinking among nulliparous women, while multiparous women at this age were as likely to binge drink in the pre-recognised part of pregnancy as multiparous women aged years old (results not shown). Irrespective of whether or not the pregnancy was planned, first-time expectant mothers were more likely to binge drink in the pre-recognised part of pregnancy than multiparous women. Women who had tried for more than 12 months to get pregnant were less likely to binge drink in the pre-recognised part of pregnancy than women who became pregnant within two months of trying OR = 0.79 (95% CI: ). The lower odds of binge drinking were confined to nulliparous women with a time to pregnancy of more than 12 months, whereas multiparous with the same time to pregnancy were as likely to binge drink as women who became pregnant within two months of trying (results not shown). Single status increased the odds of binge drinking both among pregnancy-planners and nonpregnancy-planners, OR = 1.21 (95% CI: ) and OR = 1.24 (95% CI: ). Irrespective of planningstatus, smoking, alcohol consumption before pregnancy, and week of pregnancy recognition correlates with binge drinking in the pre-recognised part of pregnancy. Women of normal weight status were more likely to binge drink in the pre-recognised part of pregnancy among pregnancy planners, whereas no association were seen among nonplanners (Table 1). Characteristics of women who binge drink in the recognised part of pregnancy Binge drinking in the recognised part of pregnancy was not related to age (Table 2). First-time pregnant women were less likely to binge drink subsequent to pregnancy recognition than multiparous women OR = 0.65 (95% CI: ). Women with unplanned pregnancies were more likely to continue binge drinking in the recognised part of pregnancy OR = 1.32 (95% CI: ), whereas women with a time to pregnancy of more than 12 months were less likely to binge drink OR = 0.81 (95% CI: ). The lower odds of binge drinking subsequent to pregnancy recognition were confined to nulliparous women, whereas multiparous women with a time to pregnancy of more than 12 months were as likely to binge drink in the recognised part of pregnancy as multiparous women who became pregnant within two months of trying (results not shown). Single status and smoking were correlated with binge drinking subsequent to pregnancy recognition OR = 1.88 (95% CI: ) and OR = 2.28 (95% CI: ), respectively. Binge drinking in the recognised part of pregnancy were less common among lower grade professionals, whereas unskilled workers and women who have been unemployed for more than one year were most likely to binge drink subsequent to recognition of pregnancy. Overweight and obese women were more likely to binge drink in the recognised part of pregnancy than women who were underweight or of normal weight status. Weekly alcohol consumption before pregnancy was a strong predictor of binge drinking also in the recognised part of pregnancy. Women who became aware of their pregnancy before week 4 were more likely to binge drink in the recognised part of pregnancy compared with women in which the pregnancy was confirmed later. Finally, self-reported presence of mental disorder or neurosis was correlated with binge drinking in the recognised part of pregnancy (Table 2). Women with a high number of binge episodes in the recognised part of pregnancy were more often multiparous, unintended pregnant, singles, had been unemployed for more than one year, had a higher alcohol consumption before pregnancy, and had more often mental disorder or neurosis compared to women with only one episode of binge drinking in the recognised part of pregnancy (Table 3). Discussion Approximately one quarter of the participants in the DNBC binge drank at least once during pregnancy; most often in the unrecognised part of pregnancy. Weekly alcohol consumption before pregnancy, single status and smoking was associated with binge drinking both in the unrecognised and recognised part of pregnancy. Besides that different characteristics were associated with binge drinking before versus after pregnancy recognition. Women who binge drank in the unrecognised part of pregnancy seemed to be somewhat more advantaged, since binge drinking in this phase of pregnancy was more common among nulliparous women aged years, and well educated with good jobs. Whereas, women who binge drank once the pregnancy was recognised are more likely to be unintended pregnant, multiparous, overweight or obese, have a mental disorder or neurosis, being an unskilled worker, or unemployed for more than one year. The proportion of binge drinkers in the DNBC is substantially lower than the 50% found in a sample of nearly 400 pregnant women seeking antenatal care in Aarhus, Denmark [13]. The reasons for differences in the proportion of binge drinkers could be that binge drinking is place and time sensitive, or that different forces of selection apply to the two studies. The Aarhus sample was recruited in 1998, which is almost in the middle of the DNBCs recruitment period, and therefore the discrepancy seems 123

76 Characteristics of pregnant binge drinkers 569 Table 2 Characteristics of women who binge drink in the recognized part of pregnancy Binge drinking in recognized part of pregnancy N % Binge drinkers OR (95% CI) a Total population 85, Age in years \ ( ) , ( ) , (Reference) , ( ) C35 9, ( ) Prior reproductive history First-time pregnant women 29, ( ) Nulliparous, history of induced abortion 4, ( ) Nulliparous, history of spontaneous abortion 5, ( ) Multiparous 45, (Reference) Time to pregnancy (months) Unplanned pregnancy 9, ( ) B2 37, (Reference) , ( ) , ( ) [12 11, ( ) Marital status Married/cohabiting 83, (Reference) Single 1, ( ) Smoking during pregnancy Smoker 21, ( ) Non-smoker 63, (Reference) Occupational status Higher grade professionals 7, ( ) Lower grade professionals 23, (Reference) Skilled workers 16, ( ) Unskilled workers 21, ( ) Students 11, ( ) Unemployed [1 year 4, ( ) Weight status before pregnancy (BMI) Underweight (\18,5) 3, ( ) Normal (18,5 24) 57, (Reference) Overweight (25 29) 16, ( ) Obese (C30) 7, ( ) Weekly alcohol consumption prior to pregnancy in drinks 0 11, (Reference) 1 27, ( ) , ( ) 4+ 25, ( ) Week of recognition of pregnancy \4 5, ( ) , (Reference) , ( ) C8 7, ( ) Self-reported mental disorder or neurosis Yes 6, ( ) No 79, (Reference) a Odds ratios from a multivariate model including all the characteristics in the table 123

77 570 K. Strandberg-Larsen et al. Table 3 Characteristics of binge drinkers in the recognized part of pregnancy according to number of episodes 1 Binge episode (N = 2,312) 2 3 Binge episodes (N = 607) 4 + Binge episodes (N = 80) Age in years (means) 29.5 (4.7) 30.0 (4.7) 29.8 (4.0) Nulliparous (%) Unintended pregnancy (%) Single (%) Smoker (%) Unemployed [1 year (%) Pre-pregnancy BMI (mean) 23.9 (4.3) 23.6 (4.2) 23.9 (4.6) Weekly alcohol consumption prior 4.2 (4.4) 5.7 (6.6) 6.9 (5.5) to pregnancy (mean) Week of recognition of pregnancy (mean) 5.2 (1.8) 5.0 (1.6) 5.1 (1.9) Self-reported mental disorder or neurosis not to be attributable to a time difference [13, 15, 16]. Around 30% of all Danish women who were pregnant during the recruitment period were enrolled into the DNBC. About half of the non-participation was caused by a lack of participation by the general practitioners, whereas the other 60% was attributable to pregnant women who declined the invitation [15]. The decision to participate may correlate with social, educational, and health conditions which again may correlate with risky health behaviour during pregnancy and it has been shown that participants in the DNBC were somewhat healthier than all pregnant mothers in Denmark [17]. However, the differential participation in the DNBC has been found to be modest and to cause very little if any bias in studies based upon internal comparisons within the cohort [17]. The potential for bias in our study is, however, a bigger issue since it is possible that the decision to participate is correlated with both binge drinking as well as other of the identified characteristics. Still, the majority of pregnancies occur among women alike the participants in the DNBC and this is presumably a group of women that health care providers pay less attention to because these are classified as low risk pregnancies, and even among this kind of women who presumably represents the healthiest: one out of four have had a binge drinking episodes while being pregnant as well as 3.5% of the women continue to binge drink after recognition of pregnancy. This proportion of pregnancy-aware binge drinkers is similar to the approximately 4% of US women who are estimated to binge drink after recognition of pregnancy [18]. As well as the identified characteristics in our study are consistent with those seen in other surveys that did not distinguish the pre- and post-recognised part of pregnancy [2, 4, 11 13, 19 21]. Nulliparous women were more likely to binge drink in the pre-recognised part of pregnancy than multiparous women. This was both the case for intended and unintended pregnancies and this implies that pregnancy-planners does not alter their drinking habits much, before confirmation of pregnancy. However, once pregnancy recognition occurred, women having their first child were those least likely to report binge drinking, as well as women with a high number of binge episodes in the recognised part of pregnancy were more often multiparous than women with only one episode in the recognised part of pregnancy. In light of women s awareness of the negative effects of alcohol on the ability to get pregnant as well as pregnancy outcome [22], it is not surprising that women with a time to pregnancy of more than one year were less likely to binge drink in the pre- and post recognised part of pregnancy than women with a shorter time to pregnancy. Contrary to our results occupational status, socioeconomic position, and education have not been found to be associated with binge drinking during pregnancy in previous studies [13, 19, 20]. The existing results concerning age are contradicting [19, 20, 23]. In this study, the associations between age and binge drinking were different for nulliparous and multiparous women as well as for binge drinking in the un- versus recognised part of pregnancy and none of the previous studies have stratified the analyses according to these factors. Our results is further supported by the fact that one of the studies that showed that women who engaged in binge drinking were older than non-binge drinking women was restricted to parous women and focused on binge drinking after recognition of pregnancy [23]. A number of study limitations are worth mentioning. Because less than 4% of all participants reported binge drinking in the recognised part of pregnancy, we had a relatively small number of persistent binge drinkers available for multivariate analyses. Moreover, the heaviest (binge) drinkers may neither participate in the DNBC nor admit the actual alcohol consumption and/or number of binge episodes. Unfortunately, no biomarkers exist that is able to provide information on how many times and when in pregnancy binge drinking had occurred and our study is therefore based on self-reported information on binge drinking. These self-reports may be affected by the 123

78 Characteristics of pregnant binge drinkers 571 perceived social desirability of a negative or attenuated response to questions on agreed health hazards, such as binge drinking [5]. Approximately three quarters of pregnant Danish women regards binge drinking in pregnancy as a potential health hazard [22]. Lack of recall may also have affected the quality of the information on binge drinking. In a previous study we showed that the reports on binge drinking were affected by the length of the recall period and information obtained in early pregnancy resulted in a higher proportion of binge drinkers [24]. Around one-third of the participants in the DNBC were interviewed in the second half of pregnancy and may have forgotten actual episodes of binge drinking. However, compared with other studies, the information in the DNBC is obtained early in pregnancy and our study also includes pregnancies that do not result in live born infants. Interviews have been shown to be a reliable method to obtain information on alcohol intake among pregnant Danish women [25] and the questions on binge drinking have been shown to yield valid and reliable information, and is the only validated method for the collection of data on timing of binge drinking [16, 26]. The number of potential risk factors may seem rather small in our study, and factors such as ethnicity, religion and use of prenatal care are lacking. The participants in the DNBC have uniform access to free and comprehensive health care and are presumed to be predominately of Danish origin and thereby Caucasians. While significant reduction in binge drinking were reported by women in the DNBC once the pregnancy was confirmed, many had experienced a binge drinking episode during the pre-recognised part of pregnancy, which is of concern since the foetus is suspected to be particularly vulnerable in the early phase of pregnancy [27]. Therefore, prevention of binge drinking in early pregnancy should not only target pregnant women, but also women at risk of becoming pregnant. This will require preventive activities targeted pregnancy-planners. The general practitioners could disseminate education on alcohol use to pregnancyplanners, when discussing birth control and sexual activity. Else, this information could be included in oral contraceptive packets. Health care providers should pay special attention to women at highest risk and this may vary from time to time and from place to place. However, we expect that some factors will play a role in most places such as social problems and/or mental problems, previous unhealthy behaviours, such as smoking indicate higher odds of binge drinking, especially if more than one characteristic is present. Acknowledgements The Danish National Research Foundation has established the Danish Epidemiology Science Centre that initiated and created the DNBC. The cohort is furthermore a result of a major grant from this foundation. Additional support for the DNBC is obtained from the Pharmacy Foundation, the Egmont Foundation, the March of Dimes Birth Defects Foundation, and the Augustinus Foundation. Sources of support: Grants from the Danish Ministry of Health and the Health Insurance Foundation. Keypoints Binge drinking is common in the pre-recognised part of pregnancy. Women who binge drink before versus after pregnancy recognition have different social characteristics. Binge drink in the pre-recognised part of pregnancy is more common among first-time mothers, aged years, who are educated or skilled workers. Binge drinking subsequent to pregnancy recognition is more common among women who are unintended pregnant, multiparous, have a mental disorder or neurosis, in unskilled works or have been unemployed for more than one year. Advisories on binge drinking in pregnancy should in addition to pregnant women be targeted pregnancyplanners. References 1. Babor T, Caetano R, Casswell S, Edwards G, Giesbrecht N, Graham K, et al. Alcohol: no ordinary commodity research an public policy. 1st ed. New York: Oxford University Press; Naimi TS, Brewer RD, Mokdad A, Denny C, Serdula MK, Marks JS. Binge drinking among US adults. JAMA. 2003;289(1): Ingersoll KS, Ceperich SD, Nettleman MD, Karanda K, Brocksen S, Johnson BA. Reducing alcohol-exposed pregnancy risk in college women: initial outcomes of a clinical trial of a motivational intervention. J Subst Abuse Treat. 2005;29(3): doi: /j.jsat Kuntsche E, Rehm J, Gmel G. Characteristics of binge drinkers in Europe. Soc Sci Med. 2004;59(1): doi: /j.socs cimed Gladstone J, Nulman I, Koren G. Reproductive risks of binge drinking during pregnancy. Reprod Toxicol. 1996;10(1):3 13. doi: / (95) Streissguth AP, Barr HM, Sampson PD. Moderate prenatal alcohol exposure: effects on child IQ and learning problems at age 7 years. Alcohol Clin Exp Res. 1990;14(5): doi: /j tb01224.x. 7. Jacobson JL, Jacobson SW, Sokol RJ, Ager JW Jr. Relation of maternal age and pattern of pregnancy drinking to Functionally significant cognitive deficit in infancy. Alcohol Clin Exp Res. 1998;22(2): doi: /j tb03659.x. 8. Bailey BN, Delaney-Black V, Covington CY, Ager J, Janisse J, Hannigan JH, et al. prenatal exposure to binge drinking and cognitive and behavioral outcomes at age 7 years. Am J Obstet Gynecol. 2004;191(3): doi: /j.ajog Royal College of Obstetricans and Gynaecologists (RCOG). Alcohol consumption and the outcomes of pregnancy. RCOG Statement No. 5-March Kesmodel U. Alcohol in pregnancy. Ugeskr Laeger. 1999; 161(36): [Article in Danish]. 11. Ebrahim SH, Luman ET, Floyd RL, Murphy CC, Bennett EM, Boyle CA. Alcohol consumption by pregnant women in the 123

79 572 K. Strandberg-Larsen et al. United States during Obstet Gynecol. 1998;92(2): doi: /s (98) Ebrahim SH, Diekman ST, Floyd RL, Decoufle P. Comparison of binge drinking among pregnant and nonpregnant women, United States, Am J Obstet Gynecol. 1999;180(1 Pt 1):1 7. doi: /s (99) Kesmodel U, Kesmodel PS, Larsen A, Secher NJ. Use of alcohol and illicit drugs among pregnant Danish women, Scand J Public Health. 2003;31(1):5 11. doi: / Floyd RL, Decoufle P, Hungerford DW. Alcohol use prior to pregnancy recognition. Am J Prev Med. 1999;17(2): doi: /S (99) Olsen J, Melbye M, Olsen SF, Sorensen TI, Aaby P, Andersen AM, et al. The Danish national birth cohort its background, structure and aim. Scand J Public Health. 2001;29(4): Kesmodel U. Binge drinking in pregnancy frequency and methodology. Am J Epidemiol. 2001;154(8): Nohr EA, Frydenberg M, Henriksen TB, Olsen J. Does low participation in cohort studies induce bias? Epidemiology. 2006;17(4): doi: /01.ede Floyd RL, Sidhu JS. Monitoring prenatal alcohol exposure. Am J Med Genet. 2004;127C(1): Naimi TS, Lipscomb LE, Brewer RD, Gilbert BC. Binge drinking in the preconception period and the risk of unintended pregnancy: implications for women and their children. Pediatrics. 2003; 111(5 Part 2): Gladstone J, Levy M, Nulman I, Koren G. Characteristics of pregnant women who engage in binge alcohol consumption. CMAJ. 1997;156(6): Flynn HA, Marcus SM, Barry KL, Blow FC. Rates and correlates of alcohol use among pregnant women in obstetrics clinics. Alcohol Clin Exp Res. 2003;27(1): Kesmodel, U, Schioler, Kesmodel P. Drinking during pregnancy: attitudes and knowledge among pregnant Danish women, Alcohol Clin Exp Res. 2002;26(10): Pascoe JM, Kokotailo PK, Broekhuizen FF. Correlates of multigravida women s binge drinking during pregnancy. A longitudinal study. Arch Pediatr Adolesc Med. 1995;149(12): Strandberg-Larsen K, Nybo Andersen AM, Olsen J, Rod Nielsen N, Grønbæk M. Do women give the same information on binge drinking during pregnancy when asked repeatedly? Eur J Clin Nutr. 2006;60: doi: /sj.ejcn Kesmodel U, Olsen SF. Self reported alcohol intake in pregnancy: comparison between four methods. J Epidemiol Community Health. 2001;55(10): doi: /jech Kesmodel U, Frydenberg M. Binge drinking during pregnancy is it possible to obtain valid information on a weekly basis? Am J Epidemiol. 2004;159(8): Allebeck P, Olsen J. Alcohol and fetal damage. Alcohol Clin Exp Res. 1998;22(7 Suppl):329S 32S. doi: /j tb04387.x. 123

80 P A P E R I I I Strandberg-Larsen K, Nielsen NR, Grønbæk M, Andersen PK, Olsen J, and Nybo Andersen AM: Binge drinking in pregnancy and risk of fetal death. Obstetrics & Gynecology 2008; 111:

81

82

83

84

85

86

87