Timing of prenatal starvation in women and birth weight in their first and second born offspring: the Dutch famine birth cohort study

European

Journal

of Obstetrics

& Gynecology

and Reproductive Biology 61 (1995) 23-30

Timing of prenatal starvation in women and birth weight in their first and second born offspring: the Dutch famine birth cohort study L.H. L~rney*“~~‘~, Aryeh D. Steinb.‘, Anita C.J. Ravellib “Deparfmenw of Obsterric.~/Gynecology and Neonatology. Acodetnic Medical Center, Amsterdam, The Netherlunds bDepartmenf of Clinical Epidemio1og.v and BiostaWics, Academic Medical Center. Amsterdam, The Netherlands CThe Gertrude H. Sergievsky Center, Faculty of Medicine. Columbia University, New York, NY. USA

Abstract Objectives: To examine the long-term effects of severe in utero maternal undernutrition on offspring birth weight. Study design. Birth weights were analyzed of 575 first born and 454 second born offspring of 683 women born in Amsterdam, the Netherlands, at the time of a severe famine at the end of World War II. In utero maternal undernutrition was defined separately for each pregnancy trimester by an average daily nutrition ration (supplied to the grandmother) of less than 1000 calories in that trimester of pregnancy. Results: Compared LO controls, birth weights of first born infants of women prenatally exposed in the first trimester of pregnancy were 73 g heavier (95% Cl: -64, 210), and birth weights of second born infants were 96 g lighter (95%) CI: -249, 58). Birth weights of infants of women exposed in the second or third trimester were much closer to controls. Conclltsions: A substantial (200 g or more) impact of severe in utero maternal undernutrition on OBW can be ruled out. There may, however, be parity specific, moderate (50-100 g) effects of maternal undernutrition early in pregnancy on OBW. This suggestion requires confirmation in other populations.

Keywords: Prenatal exposure delayed effects; Nutrition;

The

Netherlands;

the famine, per capita food availability was progressively reduced, from pre-famine levels of approximately 1600kilocalories per day to 1400kilocalories per day by January 1945, and under 1300 kilocalories per day by April 1945(Fig. 1). Official rations, which by the end of the famine consisted almost exclusively of bread and potatoes, declined much more rapidly. The severity of the famine and its widespread nature have been fully documented [l-3]. While pregnant women were allocated some additional food rations over and above those available to non-pregnant women, the extent to which redistribution of theseadditional rations occurred within families is not known. The famine ceased immediately with liberation in May 1945,when Allied food supplies became abundant. The circumstancesof the famine provide a unique opportunity to examine the long-term effects of severe in utero maternal under-nutrition on offspring birth

1. Introduction The Dutch Famine Birth Cohort Study is a historical birth cohort study of the long-term reproductive effects of starvation in utero among a cohort of women born in Amsterdam before, during and after the Dutch Hunger Winter of 1944-1945. This severe famine of 7 months duration affected the western Netherlands and occurred in a society with a well-developed administrative structure, where food supplies had been generally adequate. The famine was the result of an embargo on transport of food supplies, imposed by the German occupying forces in reprisal for a wave of partisan activity. During l Corresponding author, American Health Foundation, Epidemiology Division, 320 East 43rd Street, New York, NY 10017, USA, Tel.: 212 551 2546; Fax: 212 687 2339.

’ Present

address:

Program

in Epidemiology.

College

of Human

Medicine, Michigan State University, East Lansing, Ml, USA 0028-2243/95/$09.50 0 SSDI 0028-2243(95)02149-M

1995 Elsevier

Science

Ireland

Ltd.

All rights

Second World War; Birth weight

reserved

24

L.H.

Lumey

et al. /European

Journal

of Obstetrics

& Gynecology

and Reproductive

Biology

61 11995)

23-30

Home interviews. Women were traced through the national population registers. Complete details are provided elsewhere [6]. In the course of home interviews conducted between 1987 and 1991, information about the woman’s health and reproductive history (including date and place of birth, sex, birth weight, crown to heel length and gestational age of all her offspring), educational attainment, weight at about 18 years of age, smoking history during each pregnancy, and height was collected. All interviews were performed by a single trained research nurse and lasted l-l.5 hours.

3cOO 2600 2600 2400 2200 2000 I800 1600 I400 I200 IO00

2.2. Study population

BOO 600 400 200

Fig. 1. Average quarterly Netherlands, 1941-1945.

distribution of food rations in the western Source: Burger et al., 1948.

weight (OBW). In a previous study of birth weights of infants born at the University of Amsterdam teaching hospital between 1960 and 1984 [4], we reported a 150 g decrease in first born OBWs after maternal early pregnancy in utero exposure to the famine, and no reduction in OBWs after maternal late pregnancy in utero exposure. In that study, information regarding each mother’s perinatal experience was limited to mother’s place and date of birth and medical records of the mother’s birth were not available. The present historical birth cohort study was set up to confirm these findings, using the birth records of all women born in the University teaching hospital between August 1, 1944 and April 15, 1946. Surviving women were located at their present address and interviewed to obtain birth weights of all their offspring. We report here on our findings, relating OBW in first born and second born to the timing of maternal in utero famine exposure. Effects of the famine on pregnancy weight gain and infant birth size in the birth cohort have been previously documented [5,6]. 2. Materials

and methods

2.1. Data sources Hospital records. Obstetric records of females born

between August 1, 1944 and April 15, 1946 at the Department of Obstetrics and Gynecology, Wilhelmina Gasthuis Hospital, Amsterdam, were identified. Information about these births was collected from hospital administrative and medical records that provide, inter alia, personal identifiers, date of birth, and birth weight.

We have previously reported that, among singleton offspring, estimates of birth weight obtained by maternal recall in this population are unbiased, except among stillborn and non-surviving infants [7]. We therefore included in this analysis all mother-child pairs that fulfil the following conditions: both mother and child were singletons, the child was alive at the time of the interview, and the mother was able to recall her child’s birth weight. The analysis was limited to first and second born offspring. These were analyzed separately to avoid within-sibship correlations due to shared maternal exposure. 2.3. Famine exposure

Based on records of official food rations, the degree and timing of prenatal exposure to famine in each trimester has been documented for all monthly cohorts born in Amsterdam between August, 1944 and April, 1946 [2]. For all subjects, the date of conception was estimated by assuming that the duration of pregnancy was 9 calendar months. Since the length of gestation during the famine was reduced by 4 days on average [2], this estimate is likely to be sufficiently accurate for our purpose. The calendar month of birth was numbered 10, the preceding month 9, and so on through 1, the month of the last menstrual period. Prenatal famine exposure was defined separately for trimesters 1, 2 and 3 as an average official ration of less than 1000 calories (42005) daily in that trimester. The mean ration in the first trimester was the average daily ration in months 1, 2 and 3; for the second trimester the ration in months 4, 5 and 6; and for the third trimester the ration in months 7, 8 and 9. Using this definition, the cohorts severely exposed prenatally to famine in the first trimester were born from August to December 1945; cohorts exposed in the second trimester were born from May to September 1945; and cohorts exposed in the third trimester from February to June 1945. All other monthly birth cohorts were defined as unexposed in those trimesters. Subjects born in May-June 1945 were severely exposed in both the second and the third, and subjects born in August-September 1945 were exposed in both the first and the second trimester.

L.H.

Lumey

ef al. /European

Journal

of Obstetrics

2.4. Statistical analysis Maternal and offspring determinants of OBW (trimester of prenatal famine exposure, adult height, weight at age 18, education at time of interview, age at delivery, smoking during pregnancy, home or hospital birth and gender) were compared in first and second born infants, using unpaired t-tests for continuous variables and chisquared tests for categorical variables. Then, the association of these determinants with birth weight was compared using analysis of variance for categorical variables and t-tests for continuous variables. Lastly, the relative contribution of famine exposure to OBW, adjusting for the other determinants, was assessed in multiple linear regression models. Mother’s famine exposure status, age and pregnancy smoking status, and child’s gender were entered as categorical variables, and mother’s adult height and weight at age 18 as continuous variables. All analyses were stratified by offspring birth order. Based on our previous findings [4], we carried out a study with adequate power to detect a difference of 175-200 g in OBWs after each trimester of maternal in utero famine exposure. Statistical analyses were performed using SPSS/PC+ version 4.0 (SPSS Inc, Chicago, IL) and EpiInfo version 5 (Centers for Disease Control, Atlanta GA). The study was approved by the Medical Ethics Committee of the Academic Medical Center and the Institutional Review Board of the Columbia-Presbyterian Medical Center.

3. Results 3.1. Study population Between August 1, 1944 and April 15, 1946, 1025 liveborn singleton female infants were born in the Wilhelmina Gasthuis hospital in Amsterdam, the Netherlands. All the mothers were white and nearly all had been born in the Netherlands. Follow-up status in 1987 (died, emigrated, or currently resident in the Netherlands) was confirmed for all subjects. Of the cohort, 125 (12%) had died, and a further 87 (10% of the survivors) had emigrated from the Netherlands. Home interviews were completed for 84% (683/813) of all surviving subjects resident in the Netherlands. Of the 683 women interviewed, 72 (11%) had no children. Among first born children, 36 of 611 births (6%) did not meet the inclusion criteria for this analysis (i.e. surviving singletons with birth weight recalled by their mother). Among second born children, 34 of 488 births (7%) were excluded for the same reason. Children of birth order three and over (15% of infants) were not analyzed because of the small numbers available for study. The study population consists of 575 mother-firstborn child pairs and 454 mother-second born child pairs (Table 1). There were 437 sets of women with both a first born and a second born infant.

& @nerology

and Reproductive

Biology

61 (1995)

23-30

25

Table 1 Study population: 1025 liveborn singleton females born in the Wilhelmina Gasthuis, Amsterdam, between August I, 1944 and April I5 1946, by follow-up status and obstetric history. Dutch Famine Birth Cohort Study. n All births Died % of cohort Emigrated o/u of survivors Refused % of survivors Interviewed %I of survivors Without children “/u of interviewees First child ineligible** % of parous interviewees First child included “XI of parous interviewees Two or more children ‘I/u of interviewees Second child ineligible*** “%Jof multiparous interviewees Second child included % of multiparous interviewees

1025 (100) 125* (12.2) 87 (9.7) 130 (14.4) 683 (75.9) 72 (10.5) 36 (5.9) 515 (94. I) 488 (71.4) 34 (7.0) 454 (93.0)

‘Birthweight data missing for two births - one died in infancy and one ws not interviewed. **First-born child was multiple (n = 6), stillborn (n = 6) died prior to the date of the mother’s interview (n = 18) or maternal recall of birth weight was missing (n = 13). Some children met more than one exclusion criterion. ***Second-born child was multiple (n = 14), stillborn (n = 3). died prior to the date of the mother’s interview (n = 12) or maternal recall of birth weight was missing (n = 13). Some children met more than one exclusion criterion.

3.2. Selected characteristics offirst and second born offspring There were no differences between the first and second born offspring with respect to the mother’s prenatal exposure status. The mean birth weight in second born offspring was 95 g higher (95% CI: 21, 169) than in firstborn: 3411 g (SD. 593) vs. 3316 g (S.D. 612). Second born offspring were more likely to be born at home than first born offspring (40 vs. 24%), and their mothers were, on average, 2.5 years older than mothers of firstborn. Adult height, weight at age 18, education, prevalence of smoking during pregnancy or the child’s gender did not differ by offspring parity (Table 2). 3.3. Determinants of OBW Average birth weights in first born infants by mother’s month of birth are given in Fig. 2. Compared to the 303 control mothers not exposed to famine at all, whose infant’s mean birth weight was 3312 g, the average birth

Table 2 Selected charactertstics born offspring and Study.

of singleton surviving of their mothers. Dutch

First-born (II = 575)

Mother’s prenatal I st trimester

exposure

status 19.0

2nd trtmester 3rd trimester Controls” Child‘s Mean

birth

weight

of birth

Mother’s Mean

adult (SD.)

Mother’s Mean

weight (S.D.)

at age

completed

high

height

52.7

51.5

0.76

(612)

iJI1

(5931

0.0 I

40.3 5’) 7

< 0.00 I

23.7

(4.1)

165.2

(6.3)

165.0

(6.3)

0 61

76 1 (3 Y)

0.42

(X.9,

58.X

(X.4)

I .oo

IH (kg) 5X.X

during

Yes

school

(” ,) 52 3

52 6 47.4

O.Y7

47.7 pregnancy (“,L) 6? 3 37.7

67. I 17.9

I .oo

51.x

49.X

0.57

4x.1

50 2

(‘I/;$)

difference exposed

First

0.85 0.72 0.57

(cm)

NO

*For **Not

IX 3 23.6 24 0

(years)

Yes

Child’s gender Male Female

l

(‘% of \amplet

‘3.X 16.’

smoked

weight of the 109 infants of mothers prenatally exposed during the first trimester of pregnancy was 3390 g (difference compared to controls 78 g; 95% CI: -57, 213), of the 129 infants of mothers exposed during the second trimester 3318 g (difference compared to controls 6 g; 95% Cl: - 125, 137), and of the 131 infants of mothers exposed during the third trimester 3271 g (difference compared to controls 41 g; 95% CI: -166, 84) (Table 3). In a subgroup analysis restricted to first born infants of women exposed during the first trimester only, the average birth weight of the 28 infants without a younger sib was 303 g lower (95% CI: -560, -47) than the average birth weight of the 81 infants with a younger sib (3165 vs. 3468 g). In women exposed during the second or third trimester, there were no differences in OBW by sibship size. Average birth weights in second born infants by mother’s month of birth are given in Fig. 3. Compared to the 234 control mothers not exposed to famine at all, whose infant’s mean birth weight was 3416 g, the average birth weight of the 83 infants of mothers prenatally exposed during the first trimester of pregnancy was 3344 g (difference compared to controls 72 g; 95% CI: -200, 56), of the 107 infants of mothers exposed during the second trimester 3454 g (difference compared to controls 38 g; 95% CI: -94, 170), and of the 109 infants of mothers exposed during the third trimester 3462 g (difference compared to controls 46 g; 95% Cl: -94, 186) (Table 3). In the above analyses, the unadjusted trimester-specific effects of maternal exposure on OBW are compared to controls not exposed to famine at all, ignoring simultaneous exposures in other trimesters. The differences in OBW by maternal famine exposure status were not statistically significant. OBW was strongly correlated with mother’s adult height, weight at age 18, pregnancy smoking and child’s gender. OBWs did not vary by education and offspring’s place of birth (home versus hospital).

aecondCohort

(‘!J,I)

age at delivery (S.D.)

Mother

P

‘2.4 22.x

3316

Mother’s Mean

Mother No

Second-born (0 = 454)

and Birth

(g)

(S.D.)

Child’s nl r ace Home Hospital

first-born Famme

Infant

between lirst born to famme at all.

birth

weight

and

second

born

(g)

3600

3400

3.4. Determinants of OBW: multiple linear regression models Further adjustment of the relation of OBW to maternal exposure was made for all trimesters of in-utero exposure taken together and for the variables which showed an association with OBW. These include mother’s adult height, weight at age 18, age at delivery. pregnancy smoking status and child’s gender, and all trimesters. Compared to the control mothers, unexposed in that particular trimester, the average birth weight of first born infants of mothers prenatally exposed in the first trimester was 73 g heavier (95% CI: -64, 210), of infants born of mothers exposed in the second trimester, 10 g lighter (95% CI: -135, 115) and of infants born of mothers exposed in the third trimester, 44 g lighter (95% CI: -166, 78). Compared to controls, the birth weight of second born infants of mothers prenatally exposed in

1111111111111 3000 3200

10’44

,?45

Maternal

Cl45

month

7i45

and

10,45

year

,146

4146

of birth

Fig. 2. Average birth weight of first Infant by mothers‘ month 01 hlrth. August 1944-April 1946, University Hospital Amsterdam (n = 575)- ? monthly smoothed averages. average standard error of monthly mean 114 g.

L.H.

Table 3 Offspring

birth weight

Lumey

ef al. / European

by maternal

prenatal

Journal

famine

of Obsrerrics

exposure,

& Gynecology

adult height and weight,

First-born children Birth weight (g)

Mother’s prenatal exposure status 1st trimester 2nd trimester 3rd trimester Controls** Mother’s adult height Below 160 cm 160-164.9 cm 165-169.9 cm 170 cm and over Mother’s weight at age 18 less than 50 kg 50.0-59.9 kg 60.0-69.9 kg 70.0-79.9 kg 80 kg and over Mother completed high school No Yes Mother’s age Under 20 years 20-24 years 25-29 years 30-34 years 35 years or over Mother smoked in pregnancy No Yes Child’s place of birth Home Hospital Child’s gender Male Female

P’

ond Reproductive

Biology

and other characteristics.

61 (1995)

Dutch

23-30

Famine

Second-born Birth weight

children (g)

n

Mean

(S.D.)

83 107 109 234

3344 3454 3462 3416

(454) (541) (601) (640)

85 122 153 94

3301 3297 3479 3549

(556) (546)

54 177 175 39 9

3276 3359 3468 3594 3339

(577) (551)

27

Birth Cohort P*

n

Mean

(SD.)

109 129 131 303

3390 3318 3271 3312

(602) (63% (608) (621)

107 149 192 127

3136 3220 3342 3541

(615) (582) (612)

72 231 209 48 IS

3111 3271 3393 3399 3664

(582)

(530)

301 274

3289 3346

(622) C-501)

0.27

239 215

3399 3425

(605) (581)

0.63

99 291 140 35 IO

3369 3312 3338 3130 3238

(65% (584) (627)

0.36

20 165 204 51 14

3083 3434 3457 3367 3102

(535) (582) (583) (638) (615)

0.02

358 217

3383 3206


282 172

3475 3307

(576) (608)

0.004

(633)

137 438

3260 3333

(528) (636)

0.22

183 271

3456 3381

(610)

0.18

(581)

298 277

3378 3249

(614) (604)

0.01

226 228

3510 3313

(535)

0.25 0.93 0.52

0.001

(641) (563 (650)

(580) (810) (590)

0.27 0.57 0.52 0.002

co.oon1

(578)

Study.

(579)

(666) 0.046

(621) (621) (649)

(633)


*For mother’s prenatal exposure status: specific trimester between all levels. All comparisons within given parity. **Not exposed to famine at all.

compared

the first trimester was 96 g lighter (95% CI: -249, 58), of infants born of mothers exposed in the second trimester, 90 g heavier (95% CI: -45, 224), and of infants born of mothers exposed in the third trimester, 19 g heavier (95% CI: -113, 151) (Table 4). After adjustment, each 1 cm increase in maternal adult height was associated with an increase of 17 g (95% CI: 8, 25) in birth weight of first born and 11 g (95% CI: 1, 20) in second born infants. Each 1 kg increase in maternal adult weight was associated with an increase of 8 g (95% CI: 2, 14) in birth weight of first born and 5 g (95% CI: -2, 12) in second born infants. After adjustment, maternal smoking in pregnancy was associated with a birth weight decrease of 165 g (95% CI:

to controls.

For the other variables:

simultaneous

comparison

oi difference

-265, -65) in first born and a decrease of 157 g (95% CI: -270, -48) in second born. After adjustment, male offspring were on average 95 g heavier than females in firstborn (95% CI: 2, 191) and 171 g heavier in second born (95% CI: 65,278). Compared to the other maternal age groups, offspring of mothers aged 30 or over have significantly decreased birth weights. Birth weights are especially low in women under age 20 who are delivering their second infant. In a subgroup analysis among first born infants, adjusting also for the presence of a second infant (yes vs. no), and for the interaction of sibship size with exposure period, OBWs are associated with both the trimester of maternal fetal exposure and with sibship size. After ma-

28

L.H. Second

infant

Lumey birth

et al. /European weight

Journal

of Obstetrics

(g)

& Gynecology

Biology

61 (1995)

23-30

4. Discussion

3600

In this study, we report a 73 g increase in first born and a 96 g decreasein second born OBWs of women prenatally exposed during the first trimester of pregnancy to the Dutch famine of 1944-1945. All estimates were adjusted for the independent effects of the other trimesters of exposure, for mother’s height, weight at age 18, age, pregnancy smoking, and offspring gender. This adjustment made no difference. It could be argued that adjustment of OBW for mothers’ height and weight should not be carried out becauseany effects on OBW could be mediated in part through the effects on mothers’ stores and body composition. Becausesuch adjustment made no practical difference in our data, any famine effects on OBW appear largely independent of maternal size. Our results suggestthat the effects of maternal first trimester prenatal exposure on OBW may be parity specific. The confidence intervals of the first trimester effects in first and second born infants do not completely overlap, and there is a significant interaction at the P = 0.10 level between parity and first trimester (but not with secondor third trimester) maternal exposure when the data on first born and second born OBWs are pooled. The OBWs in first and secondborn infants are highly correlated, however, and the confidence intervals on the parameter estimatesmay need to be adjusted accordingly. Biological explanations for a parity-specific effect can, for the moment, only be speculative. In addition, our subgroup analysis, not supported by an a priori hy-

3400

3200

3000

and Reproductive

ll~llllll~llll II illI 10144 ,145 4,455 T/45 10/45 1/46 4/46 Maternal

month

and

year

of birth

Fig. 3. Average birth weight of second infant by mothers’ month of birth, August 1944-April 1946, University Hospital Amsterdam (n = 454), 3 monthly smoothed averages, average standard error of monthly mean: 129 g.

ternal first trimester exposure, there is a 154 g increase (95% CI: -2, 309; P = 0.05) in OBWs of infants with a younger sib, compared to controls not exposed in the first trimester. There is no association of OBWs with other trimesters of exposure. There is also a 319 g decrease(95% CI: -601, -38; P = 0.03) in OBWs in infants without a younger sib compared to infants with a younger sib. These additional adjustments did not materially change the parameter estimates of the other variables in the model (data not shown). Table 4 Multiple regression 1944-1946. Dutch

models for offspring Famine Birth Cohort

birthweight Study

among

First-born

children

B Constant Mother’s prenatal exposure status’ 1st trimester 2nd trimester 3rd trimester Mother’s adult height (cm) Mothers’s adult weight (kg) Mother’s age Under 20 years 20-24 years 25-29 years (reference) 30 years or over Mother smoked in pregnancy Yes (ref = no) Child’s gender Male (ref = female) Variance explained (r’)

575 firstborn

and 454 second-born

offspring

Second-born 95% CI

P

**

B

2946

of selected women

born

in Amsterdam,

children l t

95% Cl

P

(-249, 58) (-45, 224) (-113, 151) ( 1, 20)

0.39

3306

73 -10 -4 17 8

(-64, 210) (-135. 115) (-166, 78) ( 8, 25) ( 2. 14)

0.51

97 -16

( -55. (-134.

0.04

249) 103)

-213

(-41 I. -15)

-165

(-265.

95

-65)

( 2, 191)

-96 90 19 II 5

0.000 I 0.008

0.001 0.05

-289 IO

(-556, (-108,

-22) 129)

-175

(-336,

-15)

-157

(-270,

-48)

171

0.09

‘Each trrmester compared to no exposure in that trimester, “For each variable. adjusted for all others.

C-2,12)

( 65. 278)

0.03 0.16 0.03

0.005 0.002

0.08 adjusted

for exposure

in other

trimester(s)

and the other

variables.

L.H.

Lumey

et al. /European

Journal

of Obstetrics

pothesis, suggests that the first trimester exposure effect on OBWs among first born infants may be different in mothers with one compared to mothers with two or more infants. This observation needs further exploration. This study had over 80% power to detect a difference of 200 g in first born OBWs of women exposed early in gestation compared to controls, and over 70% power to detect a difference of 175 g. As such differences were not observed, except in a post-hoc subgroup analysis, an OBW effect of maternal starvation in early gestation of this magnitude should probably be ruled out. For a proper evaluation of smaller differences, it might be possible to carry out additional studies in hospitals located in famine-stricken cities of the western Netherlands with medical records for the 1944-1946 period. The increase in OBWs in first born after maternal early in utero exposure, even more pronounced among first born with a younger sib, is in contrast to our previous finding. This inconsistency could be related to sampling variability or to the difference in study design. In the present study, a well-defined birth cohort of women was followed for reproductive outcomes. Considering selected characteristics of the grandmother’s pregnancy in 1944-1946, we found no differences between women successfully interviewed in the Netherlands and women who refused to participate in the follow-up study, or with women who are no longer living in the Netherlands [6]. In our previous study, first born OBWs of women delivering in the Amsterdam teaching hospital between 1960 and 1984 were classified by a woman’s place and date of birth. OBWs in second born infants were not collected. Whereas we have argued [4] that selection bias is an unlikely explanation for the previous findings, there is no direct evidence that this study population is indeed representative of a welldefined birth cohort. Because of above design considerations, we feel the present study is the standard against which other studies should be compared. OBWs were estimated by maternal recall as birth records are no longer available for about a third of all offspring. We have shown [7] that recall data in this cohort are unbiased. Specifically, recalled OBWs did not differ between first born infants with and first born infants without a younger sib, except after first trimester maternal exposure. As recall data permit the use of all offspring, the slight reduction in precision documented for recall estimates is more than compensated for by the increased sample size. The accuracy of maternal recall of OBWs in this cohort was unrelated to the women’s date of birth. We attempted to collect offspring gestational age by maternal recall and to validate this information in extant medical records. As it turns out, a meaningful analysis of gestational age outcomes was not possible because of missing information for too many subjects.

& Gynecology

and

Reproductive

Biology

61 (1995)

23-30

29

Extant medical records documenting head circumference in the offspring were also sought because there could well be long-term effects on body symmetry in the offspring, independent of birth weight. Such data were missing for too many subjects. An analysis of offspring crown to heel length at birth and the derived Ponderal index at birth (weight/height3) by maternal in utero famine exposure (data not shown) had less power than the OBW analyses and provided no evidence for a maternal birth cohort effect on these outcomes. It has been well documented that the maternal intrauterine environment affects OBW 181. Smoking in pregnancy is associated with decreased birth weights in offspring, a result of decreased fetal growth following compromised fetal circulation. The effect of maternal smoking in pregnancy in this study is large and of equal size in first born and second born offspring. Differences in the quality of the uterine circulation have also been proposed to explain the well-documented differences in birth weight between first and second born infants [8], the maternal environment being more of a limiting factor in the first compared to later pregnancies [9]. In our study, second born offspring were 95 g heavier than first born offspring. This is in agreement with most reported findings. Whereas parity differences in birth weight might also, in part, be attributable to differences in maternal size, as heavier mothers tend to bear heavier infants and maternal weight is associated with parity, maternal weight in our study refers to weight at age 18, not to weight prior to each pregnancy. Therefore, its relative effect on first and second born could not be ascertained separately. In animal [lo] and human [l l] studies, maternal size was much more important than paternal size in determining birth weight. This was attributed to the constraining effects of the maternal environment before birth. After birth, catch-up growth takes place and over time offspring assume mid-parental size. In general, the constraining effect of maternal size on OBW is larger in first born compared to later born. Our findings are concordant. As parental size is not associated with gestational age in the offspring, any associations between maternal and infant size must again be explained by the effects of the maternal intrauterine environment on fetal growth rate. As expected, the constraining effect of the maternal intrauterine environment is more pronounced in first born compared to second born offspring. In summary, our findings provide some evidence that the intrauterine environment as provided by the mother is subject to programming by the intrauterine environment as experienced by the mother. We can rule out a substantial (200 g or more) impact of maternal in utero famine exposure on OBW. The suggestion of parityspecific, moderate (So-100 g) effects of maternal in utero exposure early in pregnancy on OBW is intriguing and requires confirmation in other populations.

30

L. H. Lumey

et al. /European

Journal

of Obstetrics

Acknowledgments P.E.Treffers, Department of Obstetrics and Gynecology, and J.G. Koppe, Department of Neonatology, Academic Medical Center, University of Amsterdam provided access to the records of the former patients of their Departments in 1944-1946. The medical archives department of the Academic Medical Center in Amsterdam, the Gemeentearchief of Amsterdam, the Bevolkingsregister of Amsterdam, and the Bevolkingsregisters in the Netherlands assisted in patient location and follow-up. The generosity of Z.A. Stein and M.W. Susser as mentors and colleagues is gratefully acknowledged. We thank D.J.P. Barker, K. Godfrey and J.H.P. van der Meulen for a critical review of the manuscript and helpful comments. Finally the follow-up studies would not have been possible without the cooperation of the women of the famine birth cohorts. Presented in part at the sixth annual meeting of the Society for Pediatric Epidemiology, Keystone, Colorado, June 1993, and at the Workshop on Nutrition in Pregnancy, Catharina Ziekenhuis, Eindhoven, the Netherlands, November 1994. Supported in part by grant 1ROl -HD-26492-02, National Institutes of Health.

& Gynecology

and Reproductive

Biology

61 f 1995)

23-30

References Ill

Burger GCE, Drummond JC, Sandstead HR. Malnutrition and Starvation in Western Netherlands, September 1944 to July 1945, Parts I and II. ‘s-Gravenhage, Staatsuitgeverij, 1948. 121 Stein ZA, Susser M, Saenger G, Marolla F. Famine and Human Development: The Dutch Hunger Winter of 1944-1945. New York: Oxford University Press, 1975. der Nederlanden in de tweede [31 De Jong L. Het Koninkrijk Wereldoorlog 1939-1945. Dee1 lob. ‘s-Gravenhage, Staatsuitgeverij, 1981. in infants after maternal in 141 Lumey LH. Decreased birthweights utero exposure to the Dutch famine of 1944-1945. Paediatr Perinatal Epidemiol 1992; 6: 240-253. op de groei van de 151 Sindram IS. De invloed van ondervoeding vrucht. Ned Tijdschr Verloskunde Gynaecol 1953; 53: 30-48. 161 Lumey LH, Ravelli ACJ, Wiessing LG, Koppe JG, Treffers PE, Stein ZA. The Dutch Famine Birth Cohort Study: Design, validation of exposure, and selected characteristics of subjects after 43 years follow-up. Paediatr Perinatal Epidemiol 1993; 7: 354-367. recall of bir[71 Lumey LH, Stein AD, Ravelli ACJ. Maternal thweights of adult children: validation by hospital and well baby clinic records. Int J Epidemiol 1994; 23: 1006-1012. 181 Thomson AM. Fetal growth and size at birth. In: Barron SL, Thomson AM, eds. Obstetrical Epidemiology. London: Academic Press, 1983; 89-141. Th, Record RG. Influence of prenatal environment 191 McKeown on correlation between birth weight and parental height, Am J Hum Genet 1954; 6: 457-463. J. The maternal effects on growth and [101 Walton A, Hammond conformation in Shire horse-Shetland pony crosses. Proc R Sot London Ser B 1938; 125: 31 l-335. 1111Cawley RH, McKeown Th, Record RG. Parental stature and birth weight. Am J Hum Genet 1954; 6: 448-456.