This study examined the relationship between maternal diet during pregnancy, maternal weight gain, and birth size parameters. The study found:
1) Intake of energy and macronutrients was not significantly associated with birth size.
2) Maternal weight gain by the end of the second trimester was significantly associated with higher energy intake and intake of protein, animal lipids, and inversely with carbohydrates.
3) While maternal weight gain is strongly linked to birth size, the nutritional associations with weight gain did not similarly correlate with birth size parameters, resembling relationships between diet, cholesterol levels, and heart disease risk.
This study examined the relationship between maternal diet during pregnancy, maternal weight gain, and birth size parameters. The study found:
1) Intake of energy and macronutrients was not significantly associated with birth size.
2) Maternal weight gain by the end of the second trimester was significantly associated with higher energy intake and intake of protein, animal lipids, and inversely with carbohydrates.
3) While maternal weight gain is strongly linked to birth size, the nutritional associations with weight gain did not similarly correlate with birth size parameters, resembling relationships between diet, cholesterol levels, and heart disease risk.
This study examined the relationship between maternal diet during pregnancy, maternal weight gain, and birth size parameters. The study found:
1) Intake of energy and macronutrients was not significantly associated with birth size.
2) Maternal weight gain by the end of the second trimester was significantly associated with higher energy intake and intake of protein, animal lipids, and inversely with carbohydrates.
3) While maternal weight gain is strongly linked to birth size, the nutritional associations with weight gain did not similarly correlate with birth size parameters, resembling relationships between diet, cholesterol levels, and heart disease risk.
Diet during pregnancy in relation to maternal weight
gain and birth size P Lagiou 1,2 , RM Tamimi 2 , LA Mucci 2 , H-O Adami 2,3 , C-C Hsieh 2,4 and D Trichopoulos 1,2 * 1 Department of Hygiene and Epidemiology, School of Medicine, University of Athens, Greece; 2 Department of Epidemiology, Harvard School of Public Health, Boston, MA, USA; 3 Department of Medical Epidemiology, Karolinska Institutet, Stockholm, Sweden; and 4 University of Massachusetts Cancer Center, Worcester, MA, USA Objective: Maternal weight gain has been consistently linked to birth weight but, beyond maternal energy intake, no macronutrient has been associated with either of them. We have examined whether maternal energy-adjusted intake of macronutrients is associated with either maternal weight gain or birth-size parameters. Design: Cohort study. Setting: University hospital in Boston, USA. Subjects: A total of 224 pregnant women coming for their first routine prenatal visit. The women were followed through delivery. Interventions: None. Pregnant womens dietary intake during the second trimester was ascertained at the 27th week of pregnancy through a food frequency questionnaire. Results: Intake of neither energy nor any of the energy-generating nutrients was significantly associated with birth size. In contrast, maternal weight gain by the end of the second trimester of pregnancy was significantly associated with energy intake ( 0.9kg/s.d. of intake; PB0.006) as well as energy-adjusted intake of protein ( 3.1kg/s.d. of intake; Po10 -4 ), lipids of animal origin ( 2.6kg/s.d. of intake; Po10 4 ) and carbohydrates (5.2kg/s.d. of intake; Po10 4 ). Conclusions: Although maternal weight gain is strongly associated with birth size, the indicated nutritional associations with weight gain are not reflected in similar associations with birth-size parameters. The pattern is reminiscent of the sequence linking diet to coronary heart disease (CHD) through cholesterol: diet has been conclusively linked to blood cholesterol levels and cholesterol levels are conclusively linked to this disease, even though the association of diet with CHD has been inconclusive and controversial. Sponsorship: This study was supported in part by Grant No. CA54220 from the National Institutes of Health European Journal of Clinical Nutrition (2004) 58, 231237. doi:10.1038/sj.ejcn.1601771 Keywords: diet; pregnancy; birth weight; maternal weight gain Introduction Birth weight is an important correlate of neonatal and infant health and has been recently associated with adult onset diseases, including cardiovascular diseases (Rich-Edwards et al, 1997), non-insulin dependent diabetes mellitus (Rich- Edwards et al, 1999) and breast cancer (Potischman & Troisi, 1999). Several studies have examined sociodemographic, reproductive and anthropometric factors in relation to birth- size parameters. In particular, birth weight is higher among offspring of women 1735y old in comparison to younger and older women (Lee et al, 1988), among multiparae in comparison to primiparae (Magnus et al, 1985) and among women of higher rather than lower socioeconomic status (Spencer et al, 1999). In contrast, smoking (England et al, 2001) and coffee consumption (Eskenazi et al, 1999) during pregnancy have been linked to lower birth weight. Although maternal weight gain has been consistently linked to birth Received 5 February 2003; revised 20 March 2003;accepted 26 March 2003 *Correspondence: D Trichopoulos, Department of Epidemiology, Harvard School of Public Health, 677 Huntington Avenue, Boston, MA 02115, USA. E-mail: dtrichop@hsph.harvard.edu Guarantor: D Trichopoulos. Contributors: The study was executed by PL, RMT and LAM. The original international study was conceived by H-OA, DT and CcH, who also provided input in the analysis. All authors contributed to the interpretation of the results and the preparation of the manu- script. European Journal of Clinical Nutrition (2004) 58, 231237 & 2004 Nature Publishing Group All rights reserved 0954-3007/04 $25.00 www.nature.com/ejcn weight (Abrams & Selvin, 1995; Zhou & Olsen, 1997; Thorsdottir & Birgisdottir, 1998; Shapiro et al, 2000; Lagiou et al, 2003), the relation of energy-generating nutrients with birth size after controlling for energy-intake has not been adequately investigated. Indeed, the sequence from maternal nutrition to maternal weight gain to birth weight is not sustained by available data, except under extreme nutri- tional deprivation (Susser, 1991). Notwithstanding findings indicating that the consumption of marine omega-3 fatty acids (Olsen, 1993) and monounsaturated lipids from olive oil (Petridou et al, 1998) may be positively associated with birth weight, the prevailing view is that qualitative aspects of diet in developed countries are unlikely to be strongly related to birth weight (Mathews et al, 1999). We have examined the relation of the intake of energy and energy-generating nutrients with maternal weight gain until the 27th gestational week and with birth-size parameters in a cohort of nonpre-eclamptic pregnant Caucasian women who had singleton births after gestation lasting from 37 to 42 weeks, inclusive, in a major university hospital in Boston USA from 1994 to 1995. We have restricted our study to this group of women and newborns, because there would be too few pre-eclamptic women or premature/overmature babies to allow evaluation of possible subtle and complex interac- tions. The hypothesis under investigation is that one or more energy-generating nutrients may have differential effects on birth weight, possibly mediated through differential effects on maternal weight gain. Materials and methods Study subjects The present investigation was undertaken using data from an international prospective study on predictors of pregnancy hormones among women in Boston, USA and Shanghai, China (Lipworth et al, 1999). Since we have an inadequate nutrient database for the Chinese diet at this time, we present here data on dietary factors in relation to maternal weight gain and birth-size variables for the US women only. Between March 1994 and October 1995, 402 eligible pregnant women were identified at the Beth Israel Hospital in Boston. To be able to participate in the study, women had to be Caucasian, less than 40-y old and have a parity of no more than two. Women were not eligible if they had taken any kind of hormonal medication during the index preg- nancy, if they had a prior diagnosis of diabetes mellitus or thyroid disease, or if the fetus had a known major anomaly. A trained health professional met all pregnant women coming for their first routine prenatal visit to the collaborat- ing maternity clinic, ascertained whether the woman was eligible to participate, explained to her the objectives of the study and the requirements for participation, and obtained informed consent. The procedures followed were in accor- dance with the ethical standards for human experimentation established by the Institutional Review Board at the Harvard School of Public Health and Beth Israel Hospital. Of 402 eligible women, 77 refused to participate, nine were subsequently excluded because the index pregnancy was terminated through a spontaneous or induced abortion, two were excluded because of twin birth and 10 were lost to follow-up after the initial meeting. For the present analysis, we excluded 23 women because they had a pregnancy that lasted less than 37 or more than 42 weeks, 16 women because they had missing data for one or more of the sociodemographic or reproductive factors evaluated in this analysis and an additional 14 women who developed pre-eclampsia. Of the remaining 251 women, 224 who had adequately completed the food frequency questionnaire, so as to allow calculation of energy and macronutrient intake, were eventually included in the present study. Information concerning the nondietary aspects of the standard questionnaire administered and medical record reviewing has been given in the earlier publication (Lipworth et al, 1999). Dietary information was collected through a semiquantitative food frequency questionnaire covering an extensive list of foods and beverages, as well as information on vitamin or multivitamin supplements. This questionnaire was identical to the one used and validated in the Nurses Health Study (Willett et al, 1985). The questionnaire was mailed to the women 1 week prior to their second routine visit to the maternity clinic, which was around the 27th gestational week. It required information on their dietary patterns during the second trimester of pregnancy and it was checked for completeness by trained interviewers during the womens second visit. Birth weight, height and head circumference were measured at delivery by study collaborators, whereas mater- nal weight gain was calculated as the difference between weight measured at the 27th gestational week and prepreg- nancy weight. Statistical analysis The statistical analyses were conducted using SAS Software version 8.0 (SAS Institute, Cary, NC, USA). Following simple cross-tabulations, the data were modeled through multiple linear regression with dependent variables, alternatively, birth weight, birth height, head circumference and maternal weight gain until the 27th gestational age. Main exposure variables were intake of energy and energy-generating nutrients, and they were calculated from the dietary data using the standard software used in the Nurses Health Study (Romieu et al, 1990). Increments equal to 1 s.d. were used for the nutritional variables to allow comparability among effect estimates of these variables, which have different levels and ranges of intake. In studying the association of dietary intakes with weight gain and pregnancy outcomes, several variables with confounding potential were controlled for, namely maternal age (categorically), maternal education (categori- cally), parity (categorically), maternal height (continuously), Diet, maternal weight gain and birth weight P Lagiou et al 232 European Journal of Clinical Nutrition prepregnancy body mass index (BMI) (continuously), preg- ravid oral contraceptive (OC) use (categorically), smoking during pregnancy (categorically), exact gestational age at delivery (continuously) and gender of the baby (categori- cally). Only six women reported alcohol intake and this referred only to low frequencies and quantities; thus, alcohol intake was not a possible confounder in these data. For all women, gestational age at delivery was estimated as the exact difference between the first day of last menstruation and the date of delivery. Results Table 1 shows birth weight, length and head circumference according to maternal characteristics and gender of off- spring. Maternal characteristics were evaluated as possible confounders of the association of energy and energy- generating nutrients with birth-size characteristics. We found positive associations of maternal height, prepreg- nancy BMI, maternal weight gain and male gender with birth-size parameters. Pregravid OC use was also positively associated with birth-size parameters in this data set, whereas Table 1 Birth weight, birth length and head circumference of babies born to 224 nonpre-eclamptic Caucasian women a in Boston USA (19941995) after gestation lasting from 37 to 42 weeks, inclusive, according to maternal characteristics Birth weight (g) Birth length (cm) Head circumference (cm) Maternal characteristics N Mean (s.e.) N Mean (s.e.) N Mean (s.e.) Maternal age (y) 1824 5 3570 (160) 5 49.60 (1.54) 5 34.90 (0.29) 2529 62 3478 (62) 62 50.19 (0.29) 62 34.35 (0.20) 3034 138 3614 (39) 138 50.92 (0.20) 135 34.74 (0.15) 35+ 19 3489 (117) 19 50.15 (0.49) 19 34.11 (0.33) Maternal education High school graduate 36 3583 (77) 36 50.30 (0.42) 36 34.66 (0.24) College graduate 92 3635 (48) 92 51.05 (0.24) 90 34.70 (0.16) Higher 94 3490 (49) 94 50.32 (0.23) 93 34.46 (0.19) Parity 1 138 3569 (40) 138 50.67 (0.20) 135 34.68 (0.13) 2 86 3558 (50) 86 50.56 (0.25) 86 34.44 (0.20) Height (cm) 159 50 3473 (71) 50 50.26 (0.31) 50 34.18 (0.19) 160164 53 3526 (71) 53 50.53 (0.28) 52 34.53 (0.20) 165169 63 3599 (54) 63 50.70 (0.33) 62 34.70 (0.18) 170+ 56 3646 (58) 56 50.94 (0.32) 55 34.75 (0.30) Prepregnancy BMI (kg/m 2 ) 18 26 3468 (72) 26 50.30 (0.39) 26 34.07 (0.24) 1921 109 3530 (45) 109 50.39 (0.23) 109 34.48 (0.18) 2224 59 3681 (60) 59 51.41 (0.30) 59 35.01 (0.19) 25 30 3549 (102) 30 50.19 (0.41) 30 34.60 (0.27) Weight gain up to 27th gestational week (kg) 7 42 3451 (66) 42 50.36 (0.33) 42 34.25 (0.22) 811 97 3514 (44) 97 50.50 (0.22) 96 34.46 (0.18) 1214 34 3614 (67) 34 50.38 (0.38) 34 34.78 (0.28) 15+ 34 3787 (102) 34 51.49 (0.48) 32 35.28 (0.25) Previous oral contraceptive use Yes 172 3602 (37) 172 50.80 (0.18) 169 34.69 (0.13) No 52 3438 (57) 52 50.00 (0.26) 52 34.23 (0.21) Smoking in pregnancy Yes 11 3555 (192) 11 49.55 (1.11) 11 33.82 (0.41) No 211 3566 (32) 211 50.67 (0.15) 208 34.64 (0.11) Gender of offspring Male 114 3629 (40) 114 51.14 (0.22) 112 34.74 (0.17) Female 110 3498 (48) 110 50.07 (0.21) 109 34.42 (0.14) a The numbers do not always add up because of missing values. Diet, maternal weight gain and birth weight P Lagiou et al 233 European Journal of Clinical Nutrition no consistent effects were evident with respect to parity, smoking during pregnancy, maternal age and educational level. In Table 2, pregnant women were distributed in quartiles by intakes, alternatively, of energy, animal lipids, vegetable lipids, carbohydrates and protein. Subsequently, we calcu- lated mean differences (and standard errors) in birth weight, birth length and head circumference between the first quartile, taken as referent, and each of the subsequent quartiles. Univariate regression-derived P-values for trend by quartile groups were obtained. With respect to birth weight and birth length, no substantial or significant trend with intake of either energy or any of the energy-generating nutrients was observed. In contrast, positive associations were apparent between head circumference and intake of either energy or any of the energy-generating nutrients. The associations were significant with respect to intake of energy or protein, but they were irregular and suggestive of possible threshold effects. Interpretation of the apparent associations and their irregularities is hindered by the fact that intake of energy and energy-generating nutrients are highly corre- lated. In Table 3, the association of maternal intake of energy and energy-generating nutrients with birth-size parameters is examined. After adjustment for the possible confounders indicated in Table 1, neither energy intake nor intake of any of the energy-generating nutrients (after adjustment for energy intake) was associated with any of the birth-size parameters examined. For Table 4, complete data for 207 pairs of mothers and newborn were available, but their distributions by the variables indicated in Table 1 were similar to the distribu- tions given in Table 1. Again, pregnant women were distributed in quartiles by intakes, alternatively, of energy, animal lipids, vegetable lipids, carbohydrates and protein. Subsequently, mean differences (and standard errors) in maternal weight gain between the first quartile, taken as referent, and each of the subsequent quartiles were calcu- lated. Univariate regression-derived P-values for trend by quartile groups were obtained. With the exception of carbohydrates, significant positive associations with weight gain were evident, but the possibility of confounding, either mutual among the nutritional factors or by the nonnutri- tional factors indicated in Table 1, cannot be excluded. Table 5 shows the association of maternal intake of energy and energy-generating nutrients with maternal weight gain. After adjustment for the possible confounders indicated in the footnote of the table, energy intake and energy-adjusted intake of animal lipids and protein were significantly positively associated with maternal weight gain, whereas a significant inverse association was evident with respect to carbohydrates. In an alternative model, we introduced simultaneously the four energy-generating nutrients, but we excluded energy intake to avoid collinearity. The association of maternal weight gain with intake of carbohy- drates (inverse) and with animal lipids and protein (positive) persisted, although the partial regression coefficients were all reduced (in absolute terms). Table 2 Mean change and standard error (s.e.) in birth weight, birth length and head circumference by quartile a of intake of energy and energy- generating nutrients Q2 Q3 Q4 Q1 Mean change s.e. Mean change s.e. Mean change s.e. P-value trend Birth weight(g) Energy REF +149.0 87.9 17.1 87.9 +165.1 87.9 0.24 Animal lipids REF +132.3 88.4 4.9 88.4 +131.3 88.4 0.36 Vegetable lipids REF +65.1 88.9 4.2 88.9 +105.4 88.9 0.38 Carbohydrates REF +181.4 88.2 +51.3 88.2 +160.6 88.2 0.21 Protein REF +11.1 89.0 +79.7 89.0 +88.3 89.0 0.24 Birth length (cm) Energy REF +0.15 0.44 0.40 0.44 +0.54 0.44 0.45 Animal lipids REF 0.01 0.44 0.46 0.44 +0.22 0.44 0.87 Vegetable lipids REF 0.20 0.44 0.11 0.44 +0.17 0.44 0.66 Carbohydrates REF +0.24 0.44 +0.16 0.44 +0.30 0.44 0.56 Protein REF 0.66 0.44 0.47 0.44 +0.26 0.44 0.49 Head circumference (cm) Energy REF +0.53 0.31 +0.14 0.31 +0.92 0.31 0.02 Animal lipids REF +0.73 0.31 +0.33 0.31 +0.68 0.31 0.10 Vegetable lipids REF +0.10 0.32 0.03 0.32 +0.48 0.31 0.18 Carbohydrates REF +0.75 0.31 +0.47 0.31 +0.71 0.31 0.06 Protein REF +0.05 0.31 +0.31 0.31 +0.67 0.31 0.02 a The 25th, 50th and 75th centiles were for energy 6386, 8345 and 10146 kJ, for animal lipids 27.4, 35.1 and 47.2g, for vegetable lipids 16.9, 22.8 and 29.6g, for carbohydrates 208.2, 271.8 and 345.5 g and for protein 67.2, 86.3, and 113.4 g, respectively. Data from 224 nonpre-eclamptic pregnancies lasting from 37 to 42 weeks, inclusive, in Boston USA, 19941995. Diet, maternal weight gain and birth weight P Lagiou et al 234 European Journal of Clinical Nutrition Discussion In an earlier paper (Lagiou et al, in press), examining non- nutritional factors in relation to birth weight, we found, as others have (Abrams & Selvin, 1995; Zhou & Olsen, 1997; Thorsdottir & Birgisdottir, 1998; Shapiro et al, 2000), that maternal weight gain is positively associated with birth weight. Among Caucasian women, an increase in maternal weight gain by 2kg was associated with an increase of 37.1g (standard error 13.2g) in birth weight, after adjustment for possible confounding variables (Lagiou et al, in press). In the present paper, we found that neither energy intake nor the energy-adjusted intake of any of the energy-generating nutrients, as ascertained at the end of the second trimester of pregnancy, is significantly associated with any of the studied birth-size parameters after adjustment for confound- ing variables. In contrast, energy intake was significantly positively associated with maternal weight gain through the end of the second trimester of pregnancy and, after controlling for energy intake, protein and lipids of animal origin were also significantly positively associated with Table 3 Partial regression coefficients showing changes in birth weight (g), birth length (cm) and head circumference (cm) per increments of one standard deviation (s.d.) a of intake of energy and energy generating nutrients b Birth weight change (g) c s.e. P-value Adjusted birth weight change (g) d s.e. P-value Energy (per s.d.) +51.2 31.3 0.10 +7.8 28.8 0.79 Animal lipids (per s.d.) +6.5 48.3 0.89 +24.8 44. 5 0.58 Vegetable lipids (per s.d.) +55.1 46.7 0.24 6.9 42.2 0.87 Carbohydrates (per s.d.) 42.1 83.6 0.62 38.6 77.0 0.62 Protein (per s.d.) +3.5 68.8 0.96 +32.2 62.7 0.61 Birth length change (cm) c s.e. P-value Adjusted birth length change (cm) d s.e. P-value Energy (per s.d.) +0.15 0.16 0.34 0.06 0.14 0.66 Animal lipids (per s.d.) 0.002 0.24 0.99 +0.16 0.22 0.46 Vegetable lipids (per s.d.) +0.08 0.23 0.74 0.22 0.21 0.30 Carbohydrates (per s.d.) 0.11 0.42 0.79 0.20 0.38 0.61 Protein (per s.d.) +0.16 0.34 0.65 +0.32 0.31 0.31 Head circumference change (cm) c s.e. P-value Adjusted head circumference change (cm) d s.e. P-value Energy (per s.d.) +0.29 0.11 0.01 +0.15 0.11 0.16 Animal lipids (per s.d.) +0.01 0.17 0.95 +0.07 0.16 0.65 Vegetable lipids (per s.d.) +0.12 0.16 0.45 0.04 0.15 0.79 Carbohydrates (per s.d.) 0.13 0.29 0.65 0.16 0.28 0.56 Protein (per s.d.) +0.05 0.24 0.84 +0.16 0.23 0.47 a s.d. for energy 3371.9 kJ/day, for animal lipids 19.6g/day, for vegetable lipids 10.8g/day, for carbohydrates 127.3 g/day and for protein 40.7g/day b Total of 224 singleton, nonpre-eclamptic pregnancies, lasting from 37 to 42 weeks inclusive. Boston, USA, 19941995. c Adjusted for energy intake only (except for energy). d Adjusted for energy intake (except for energy), maternal age, maternal education, parity, maternal height, prepregnancy BMI, pregravid OC use, smoking during pregnancy, exact gestational age at delivery and gender of the baby. Table 4 Mean change and s.e. in maternal weight gain by quartile of intake of energy and energy-generating nutrients Weight gain (kg) Q2 Q3 Q4 Q1 Mean change s.e. Mean change s.e. Mean change s.e. P-value trend Energy REF +0.79 0.97 0.12 0.95 +2.46 0.96 0.04 Animal lipids REF 0.18 0.96 +0.29 0.95 +2.82 0.94 0.003 Vegetable lipids REF 0.00 0.96 0.17 0.96 +2.80 0.96 0.007 Carbohydrates REF +0.62 0.98 0.007 0.98 +1.03 0.96 0.40 Protein REF 0.14 0.96 +0.51 0.98 +1.90 0.95 0.03 Data from 207 nonpre-eclamptic pregnancies lasting from 37 to 42 weeks, inclusive, in Boston USA, 19941995. Diet, maternal weight gain and birth weight P Lagiou et al 235 European Journal of Clinical Nutrition maternal weight gain, whereas carbohydrates were signifi- cantly inversely associated with it. The strong associations between four of the five studied nutritional variables and maternal weight gain, in combina- tion with the strong positive association of the latter variable with birth weight (and, indeed, birth length and birth head circumferenceFTable 1) would have led to the prediction of significant associations between the nutritional variables and birth-size parameters. The absence of such associations is intriguing. There is a biological precedent, however, although in a different time scale. The relation of diet to coronary heart disease (CHD) has been weak in most epidemiological investigations, even though cholesterol levels (both high- and low-density lipoprotein cholesterol) are powerful predictors of CHD and are clearly associated with diet (Willett, 1998). It appears that the effects of nutritional variables are diluted by those of other determi- nants of birth-size parameters, so that they might only be detected in very large studies. Nevertheless, such effects are likely to exist and, if they are dose-dependent and exposures are extreme, they could even have physiological implica- tions. There have been earlier studies indicating that maternal protein intake is positively associated with pregnancy weight gain (Scholl et al, 1991; Kramer, 2000a) with birth weight (Weigel et al, 1991; Godfrey et al, 1997) or both (Kramer, 2000b). Other studies indicate that maternal intake of fat (Weigel et al, 1991) and carbohydrates (Godfrey et al, 1997) is, respectively, positively and inversely associated with birth weight. A study in rats showed that protein intake during pregnancy is positively associated with pregnancy weight gain and birth weight (Levy & Jackson, 1993). Finally, there have been many reports indicating that overt malnutrition and reduction in maternal energy intake are associated with reduced pregnancy weight gain and birth weight (Susser, 1991; Alexy et al, 1997; Bergmann et al, 1997; Rondo & Tomkins, 1999; Rush, 2001). Thus, the existing collective evidence is not incompatible with our findings, which, however, present a more integrated perspective. Among the advantages of our study are its prospective nature and its reliance on a validated food frequency questionnaire. Weaknesses of the study include the moderate size and the focus on pregnancy weight gain during the first two trimesters. However, several reports have indicated that maternal weight gain in the first and second trimester may be stronger determinants of newborn size than weight gain in the third trimester of pregnancy (Abrams & Selvin, 1995; Brown et al, 2002; Guihard-Costa et al, 2002). In any case, even if weight gain during the third trimester has its own determinants and consequences, this does not affect the findings of the present study. Prepregnancy weight was self- reported, but there is evidence in the literature (Yu & Nagey, 1992) that self-reported prepregnancy weight is highly correlated (rB0.9) with the objective measurement. A high proportion of women were excluded, but most of these exclusions were imposed by technical or administrative reasons that were unlikely to have introduced selection bias. The study group was not representative of the American population, but representativeness is not a prerequisite for validity in prospective cohort studies, the strength of which stems from the lack of association between errors in exposure and outcome ascertainment. In conclusion, we have found evidence that, after adjust- ment for energy intake, intake of protein and lipids of animal origin is positively associated with weight gain through the end of the second trimester of pregnancy, whereas intake of carbohydrates is inversely associated with it. Although weight gain is strongly associated with birth-size parameters, the indicated nutritional associations are not reflected in similar associations with birth-size variables. The pattern is reminiscent of the sequence linking diet to CHD through cholesterol and may be explained by the operation of non-nutritional determinants of birth size. References Abrams B & Selvin S (1995): Maternal weight gain pattern and birth weight. Obstet. Gynecol. 86,163169. Alexy B, Nichols B, Heverly MA & Garzon L (1997): Prenatal factors and birth outcomes in the public health service: a rural/urban comparison. Res. Nurs. Health 20, 6170. Bergmann MM, Flagg EW, Miracle-McMahill HL & Boeing H (1997): Energy intake and net weight gain in pregnant women according Table 5 Partial regression coefficients showing changes in maternal weight gain (kg) by the 27th gestational week per increments of 1 s.d. of intake of energy and energy-generating nutrients a Weight gain change b (kg) s.e. P-value trend Adjusted weight gain change c (kg) s.e. P-value trend Energy (per s.d.) +1.29 0.33 0.0001 +0.91 0.33 0.006 Animal lipids (per s.d.) +3.01 0.46 o0.0001 +2.56 0.47 o0.0001 Vegetable lipids (per s.d.) +1.02 0.49 0.04 +0.77 0.48 0.11 Carbohydrates (per s.d.) 5.94 0.77 o0.0001 5.22 0.80 o0.0001 Protein (per s.d.) +3.58 0.72 o0.0001 +3.11 0.71 o0.0001 a A total of 207 singleton, nonpre-eclamptic pregnancies, lasting from 37 to 42 weeks inclusive. Boston, USA, 19941995. b Adjusted for energy intake only (except for energy). c Adjusted for energy intake (except for energy), maternal age, maternal education, parity, maternal height, prepregnancy BMI, pregravid OC use, smoking during pregnancy, exact gestational age at delivery and gender of the baby. Diet, maternal weight gain and birth weight P Lagiou et al 236 European Journal of Clinical Nutrition to body mass index (BMI) status. Int. J. Obes. Relat. Metab. Disord. 21,10101017. Brown JE, Murtaugh MA, Jacobs Jr DR & Margellos HC. (2002): Variation in newborn size according to pregnancy weight change by trimester. Am. J. Clin. Nutr. 76, 205209. England LJ, Kendrick JS, Wilson HG, Merritt RK, Gargiullo PM & Zahniser SC (2001): Effects of smoking reduction during preg- nancy on the birth weight of term infants. Am. J. Epidemiol. 154, 694701. Eskenazi B, Stapleton AL, Kharrazi M & Chee WY (1999): Associa- tions between maternal decaffeinated and caffeinated coffee consumption and fetal growth and gestational duration. Epide- miology 10, 242249. Godfrey KM, Barker DJ, Robinson S & Osmond C (1997): Maternal birthweight and diet in pregnancy in relation to the infants thinness at birth. Br. J. Obstet. Gynaecol. 104, 663667. Guihard-Costa AM, Papiernik E, Grange G & Richard A (2002): Gender differences in neonatal subcutaneous fat store in late gestation in relation to maternal weight gain. Ann. Hum. Biol. 29, 2636. Kramer MS (2000a): High Protein Supplementation in Pregnancy. Cochrane Database Syst Rev, Vol. 2, CD000105. Kramer MS (2000b): Balanced Protein/Energy Supplementation in Pregnancy. Cochrane Database Syst Rev, Vol. 2, CD000032. Lagiou P, Hsieh CC, Trichopoulos D, Xu B, Wuu J, Mucci L, Tamimi R, Adami HO & Cnattingius S (2003): Birth weight differences between USA and China and their relevance to breast cancer etiology. Int. J. Epidemiol. 32, 193198. Lee KS, Ferguson RM, Corpuz M & Gartner LM (1988): Maternal age and incidence of low birth weight at term: a population study. Am. J. Obstet. Gynecol. 158, 8489. Levy L & Jackson AA (1993): Modest restriction of dietary protein during pregnancy in the rat: fetal and placental growth. J. Dev. Physiol. 19, 113118. Lipworth L, Hsieh Cc, Wide L, Ekbom A, Yu SZ, Yu GP, Xu B, Hellerstein S, Carlstrom K, Trichopoulos D & Adami HO (1999): Maternal pregnancy hormone levels in an area with a high incidence (Boston, USA) and in an area with a low incidence (Shanghai, China) of breast cancer. Br. J. Cancer. 79, 712. Magnus P, Berg K & Bjerkedal T (1985): The association of parity and birth weight: testing the sensitization hypothesis. Early Hum. Dev. 12, 4954. Mathews F, Yudkin P & Neil A (1999): Influence of maternal nutrition on outcome of pregnancy: prospective cohort study. BMJ 319, 339343. Olsen SF (1993): Consumption of marine n-3 fatty acids during pregnancy as a possible determinant of birth weight. A review of the current epidemiologic evidence. Epidemiol. Rev. 15, 399413. Petridou E, Stoikidou M, Diamantopoulou M, Mera E, Dessypris N & Trichopoulos D (1998): Diet during pregnancy in relation to birthweight in healthy singletons. Child Care Health Dev. 24, 229242. Potischman N & Troisi R (1999): In-utero and early life exposures in relation to risk of breast cancer. Cancer Causes Control 10, 561573. Rich-Edwards JW, Stampfer MJ, Manson JE, Rosner B, Hankinson SE, Colditz GA, Willett WC & Hennekens CH (1997): Birth weight and risk of cardiovascular disease in a cohort of women followed up since 1976. BMJ 315, 396400. Rich-Edwards JW, Colditz GA, Stampfer MJ, Willett WC, Gillman MW, Hennekens CH, Speizer FE & Manson JE (1999): Birthweight and the risk for type 2 diabetes mellitus in adult women. Ann. Intern. Med. 130, 278284. Romieu I, Stampfer MJ, Stryker WS, Hernandez M, Kaplan L, Sober A, Rosner B & Willett WC (1990): Food predictors of plasma beta- carotene and alpha-tocopherol: validation of a food frequency questionnaire. Am. J. Epidemiol. 131, 864876. Rondo PH & Tomkins AM (1999): Maternal and neonatal anthro- pometry. Ann. Trop. Paediatr. 19,349356. Rush D (2001): Maternal nutrition and perinatal survival. Nutr. Rev. 59, 315326. Scholl TO, Hediger ML, Khoo CS, Healey MF & Rawson NL (1991): Maternal weight gain, diet and infant birth weight: correla- tions during adolescent pregnancy. J. Clin. Epidemiol. 44, 423428. Shapiro C, Sutija VG & Bush J (2000): Effect of maternal weight gain on infant birth weight. J. Perinat. Med. 28, 428431. Spencer N, Bambang S, Logan S & Gill L (1999): Socioeconomic status and birth weight: comparison of an area-based measure with the Registrar Generals social class. J. Epidemiol. Commun. Health 53, 495498. Susser M (1991): Maternal weight gain, infant birth weight, and diet: causal sequences. Am. J. Clin. Nutr. 53, 13841396. Thorsdottir I & Birgisdottir BE (1998): Different weight gain in women of normal weight before pregnancy: postpartum weight and birth weight. Obstet. Gynecol. 92, 377383. Weigel MM, Narvaez WM, Lopez A, Felix C & Lopez P (1991): Prenatal diet, nutrient intake and pregnancy outcome in urban Ecuadorian primiparas. Arch. Latinoam. Nutr. 41, 2137. Willett WC, Sampson L, Stampfer MJ, Rosner B, Bain C, Witschi J, Hennekens CH & Speizer FE (1985): Reproducibility and validity of a semiquantitative food frequency questionnaire. Am. J. Epidemiol. 122, 5165. Willett W. (1998): Diet and coronary heart disease. In Nutritional Epidemiology, W Willett (ed). 2nd Edition, pp 414466. New York: Oxford University Press, 1998. Yu SM & Nagey DA. (1992): Validity of self-reported pregravid weight. Ann. Epidemiol. 2: 715721. Zhou W & Olsen J (1997): Gestational weight gain as a predictor of birth and placenta weight according to pre-pregnancy body mass index. Acta. Obstet. Gynecol. Scand. 76, 300307. Diet, maternal weight gain and birth weight P Lagiou et al 237 European Journal of Clinical Nutrition