Gestational bodyweight gain among underweight Japanese women related to small-for-gestational-age birth

Authors


Dr Nobuko Harita, Department of Preventive Medicine and Environmental Health, Osaka City University Graduate School of Medicine, 1-4-3, Asahi-machi, Abeno-ku, Osaka 545-8585, Japan. Email: haritan@med.osaka-cu.ac.jp

Abstract

Aim:  The prevalence of underweight women, who have an increased risk for small-for-gestational-age (SGA) birth, is increasing in Japan. We examined the associations of pre-pregnancy body mass index and gestational weight gain (GWG) with SGA birth among Japanese women.

Material and Methods:  We conducted a prospective cohort study of 1391 women who delivered full-term singleton babies. SGA was defined as below the 10th percentile of birthweight at each gestational age, baby sex, and parity. We calculated the 5th percentile of birthweight in the same way for another threshold for SGA. According to pre-pregnancy body mass index, we divided the participants into three groups: underweight (<18.5 kg/m2), normal weight (18.5–24.9 kg/m2), and overweight and obese (≥25.0 kg/m2).

Results:  SGA birth was observed most frequently among the underweight group (13.8%). Underweight was associated with an increased risk of SGA birth. The multiple-adjusted odds ratio for underweight was 1.96 (95% confidence interval, 1.23–3.11) compared with normal weight. Sufficient GWG reduced the incidence and the multiple-adjusted odds ratio for 1-kg increase of GWG was 0.86 (0.81–0.92). The same tendency was observed for the delivery of infants below the 5th birthweight percentile. Women with underweight and normal weight who had 9.0 kg or less of GWG had a significantly higher risk of SGA birth than women with normal weight who had 9.1–11.0 kg of GWG.

Conclusions:  Underweight and poor GWG were associated with a higher incidence of SGA birth. However, the incidence of SGA birth among underweight women was not increased significantly if they had sufficient GWG.

Introduction

Small-for-gestational-age (SGA) babies are known to be associated with increased perinatal mortality and morbidity.1 Pre-pregnant underweight status has been reported to be a risk factor of SGA birth,2–4 and this association is of particular concern in Japan, where about 25% of young women are underweight.5 Poor gestational weight gain (GWG) has also been reported to be associated with a greater incidence of SGA birth.2,3,6,7 Furthermore, the US Institute of Medicine (IOM) proposed guidelines for GWG using pre-pregnancy body mass index (BMI) to avoid pregnancy complications, including SGA birth, demonstrating that appropriate GWG reduces the risk for SGA birth in underweight women.8 However, as these guidelines were not intended for use with women of certain ethnic groups who are substantially shorter and thinner, they do not seem to be applicable for Japanese as well as other Asian women. Although two historical studies on Japanese women have reported the association of pre-pregnancy BMI and GWG with the risk of SGA simultaneously,3,4 the comparative importance and correlation between these two factors for the risk of SGA have not been assessed in a prospective manner.

We prospectively investigated whether pre-pregnancy BMI and GWG were independently associated with the risk of SGA birth in Japanese women. We also examined the effects of the combination of pre-pregnancy BMI and GWG on the risk of SGA birth. As infant birthweight less than the 5th percentile is considered to be more clinically meaningful with higher morbidity and mortality than SGA defined as infant birthweight below the 10th percentile (which included small infants who were not pathologically growth-restricted),8 in addition to SGA defined as infant birthweight below the 10th percentile for their gestational age,9 we evaluated the associations of pre-pregnancy BMI and GWG with the risk of delivery of infants below the 5th birthweight percentile.

Methods

Hirakata Risks Associated with Pregnancy Assessment Research (HIRAPAR) is an ongoing observational cohort study conducted at a private maternity hospital in Hirakata, Osaka, Japan, which is a city with a population of 407 418; 3554 newborns were born in the year 2009. This hospital generally manages pregnant women without major complications who are living in and around Hirakata, and refers those at possibly higher perinatal risk to a more appropriate medical center according to the patients' conditions in early pregnancy. For the HIRAPAR, we asked all pregnant women, except those who intended to interrupt their pregnancy and those who already showed clear signs of miscarriage, to complete a questionnaire in early pregnancy, usually at the first or second visit to this hospital. All pregnant women who intended to give birth at this hospital and completed the questionnaire after informed consent was provided were enrolled. The protocol for this study was reviewed by the Human Subjects Review Committee at Osaka City University.

For the current study, our subjects were women who delivered full-term singleton babies among the participants who were enrolled in this study from November 2006 to April 2008. Thus, of 1716 participants overall, we excluded 164 women because their pregnancies ended in abortion or intrauterine fetal death, and an additional 81 women who did not deliver at the hospital because they were transferred or referred elsewhere owing to medical indications. Of them, seven women were referred because their fetuses were suspected of having intrauterine growth restriction. In addition, we excluded another 54 women with twin pregnancy or preterm delivery and another 26 women because of lack of any information of covariates. Thus, the analytic cohort consisted of 1391 women.

The questionnaire elicited information on height, pre-pregnancy bodyweight, parity, and smoking and drinking habits. Information about infertility treatment was obtained from a referral letter and/or the questionnaire. Pre-pregnancy BMI was calculated as pre-pregnancy bodyweight divided by height squared. According to pre-pregnancy BMI, we divided all participants into three groups: underweight (<18.5 kg/m2), normal range (18.5–24.9 kg/m2), and overweight and obese (≥25.0 kg/m2) according to World Health Organization guidelines.10 With regard to parity, the participants were classified as primipara and multipara. Regarding smoking habit, the questionnaire had three possible answers: current smoker, past smoker, or non-smoker. Participants were classified as current smokers or non-current smokers. Those who quit smoking after conception were classified as non-current smokers. If participants answered that they consumed alcoholic drinks once or more a week, we classified them as having a drinking habit. We did not discuss details of infertility treatment here and divided the participants into two groups: no treatment or under any treatment.

We used gestational age based on the women's last menstrual periods and confirmed this by ultrasound measurement of crown–rump length, mainly at 8–11 gestational weeks. If there were discrepancies between gestational ages determined by these two methods, we adopted gestational age estimated using ultrasound measurement of crown–rump length. Among some participants in the treatment of infertility, the gestational age was determined according to its treatment schedule. Regular prenatal visits were performed basically every 4 weeks before 28 weeks of gestation, every 2 weeks before the 36th week, every week until the expected date of confinement, and twice a week thereafter. At regular prenatal visits, bodyweight was routinely measured while participants were wearing light clothing without shoes. Bodyweight was also measured during hospitalization for clinical indications other than delivery. GWG was calculated by subtraction of pre-pregnancy bodyweight from bodyweight at the last prenatal visit before delivery. Neonatal birthweight was measured soon after delivery.

SGA infants were defined as infants who had birthweights below the 10th percentile in each gestational week. As baby sex and parity influence baby birthweight, we calculated each 10th percentile of the four groups according to baby sex and parity at each gestational week. Namely, we stratified all 1417 full-term singleton infants born during the study period into these groups and then calculated each 10th percentile to define SGA birth (Fig. 1). In this study, we also calculated the 5th percentile of birthweight at each gestational week in the same four categories to examine associations of pre-pregnancy BMI and gestational weight gain with the risk of delivery of the infants below the 5th birthweight percentile (Fig. 1).

Figure 1.

Median, cut-off values for (inline image) 10th and (inline image) 5th percentile of birthweight at each gestational week, in the subgroups according to baby sex and parity. (inline image) Median.

We used multiple logistic regression models to assess the association of pre-pregnancy BMI and GWG with the incidence of SGA. Presence of effect modification was tested by insertion of first-order interaction terms into appropriate regression models. We calculated 95% confidence interval (CI) for each odds ratio (OR). The current study included a sufficient number of participants to assess the effects of the combination of pre-pregnancy BMI and GWG on the risk of SGA birth using the multiple logistic regression models. Thus, regarding the risk for delivery of infants below the 5th percentile of birthweight, the number of study subjects was insufficient to assess the effects of the combination of pre-pregnancy BMI and GWG using multiple logistic regression models. All P-values were two-tailed and considered statistically significant if the values were less than 0.05. All statistical analyses were performed using the pasw Statistics 17.0.

Results

According to our definition, there were 131 (9.4%) incident cases of SGA, of which 59 had birthweight below the 5th percentile. The characteristics of the study participants are shown in Table 1. Underweight women tended to be primiparous. The smaller the maternal pre-pregnancy BMI, the lower the birthweight of their infants. Pre-pregnancy underweight women tended to give birth to SGA infants.

Table 1.  Characteristics of the participants
 TotalPre-pregnancy category according to BMI
UnderweightNormal weightOverweight and obese
<18.5 kg/m218.5–24.9 kg/m2≥25.0 kg/m2
  1. Data are n, mean ± SD or %, as appropriate. †SGA was defined as below the 10th percentile of birthweight at each gestational age, baby sex, and mother's parity. BMI, body mass index; SD, standard deviation; SGA, small-for-gestational-age.

n 1391210108695
Mother    
 Age at entry (years)30.3 ± 4.530.1 ± 4.330.3 ± 4.630.6 ± 4.3
 Gestational weight gain (kg)9.6 ± 3.110.1 ± 2.89.7 ± 3.06.8 ± 3.8
 Height (cm)158.7 ± 5.3159.7 ± 5.1158.5 ± 5.3158.5 ± 5.8
 Pre-pregnancy BMI (kg/m2)20.9 ± 2.717.7 ± 0.720.9 ± 1.627.7 ± 2.5
 Current smoker (%)6.46.76.37.4
 Drinking habit (%)25.326.725.717.9
 Primipara (%)52.057.151.448.4
 Infertility treatment (%)13.911.013.723.2
Baby    
 Male baby (%)49.249.549.644.2
 Birthweight (g)3081.9 ± 360.23034.8 ± 376.13085.3 ± 354.93147.0 ± 376.4
SGA    
 <10th percentile (%)9.413.88.86.3
 <5th percentile (%)4.26.73.93.2

We examined the effects of pre-pregnancy BMI and GWG upon SGA birth using multiple logistic regression models (Table 2). Both pre-pregnancy BMI and GWG were independently associated with the risk of SGA birth. Pre-pregnancy underweight women were associated with an increased risk of SGA birth (OR = 1.96, 95%CI 1.23–3.11, P = 0.004) compared with those with pre-pregnancy normal weight after adjustment for GWG, maternal age at entry, short stature, smoking habit, drinking habit, and infertility treatment. Increasing GWG was associated with a lower incidence of SGA birth. The multiple-adjusted OR of GWG of SGA birth was 0.86 (95%CI 0.81–0.92) per 1-kg increase. Another anthropometric factor that increased risk of SGA birth was maternal short stature. It showed a non-linear correlation with the incidence of SGA. The incidence of SGA was greater in women who were shorter than 155 cm than in their counterparts. Therefore, we divided the participants according to height: short stature or not. The multiple-adjusted OR of short stature, that is, height less than 155 cm, was 2.31 (95%CI 1.56–3.43, P < 0.001). We examined the significance of the interaction terms between pre-pregnancy BMI, GWG, and all the other variables: maternal age at entry, short stature, smoking habit, drinking habit and infertility treatment. None of them was significant.

Table 2.  Multivariate model of the incidence of SGA birth
Independent variablesCrude OR95%CI P Multiple-adjusted OR95%CI P
  1. †SGA was defined as below the 10th percentile of birthweight at each gestational age, baby sex, and mother's parity. ‡Multiple-adjusted model was adjusted for all the variables listed here. BMI, body mass index; CI, confidence interval; OR, odds ratio; SGA, small-for-gestational-age.

Age at entry (years)      
 15–251.020.58–1.790.9541.010.56–1.810.986
 26–301.00  1.00  
 31–351.310.86–1.990.2061.330.87–2.050.190
 36–420.710.35–1.400.3190.770.38–1.560.463
Gestational weight gain (kg)0.890.84–0.94<0.0010.860.81–0.92<0.001
Pre-pregnancy BMI (kg/m2)      
 Underweight (<18.5)1.651.06–2.580.0271.961.23–3.110.004
 Normal weight (18.5–24.9)1.00  1.00  
 Overweight and obese (≥25.0)0.700.30–1.630.4030.390.16–0.960.040
Height (cm)      
 <1552.191.50–3.20<0.0012.291.55–3.40<0.001
 ≥1551.00  1.00  
Smoking habit      
 Non-current smoker1.00  1.00  
 Current smoker2.071.15–3.730.0152.941.56–5.530.001
Drinking habit      
 <1 time a week1.00  1.00  
 ≥1 time a week0.990.66–1.500.9750.900.58–1.390.626
Infertility treatment      
 No treatment1.00  1.00  
 Any treatment0.980.58–1.660.9431.020.59–1.760.937

We examined the combined effects of both pre-pregnancy BMI and GWG upon SGA birth. GWG was divided into intervals of 2.0 kg. As only a few women gained less than 7.0 kg or more than 13.0 kg during pregnancy among pre-pregnancy underweight and normal women, GWG in these groups was classified as ≤7.0 kg, 7.1–9.0 kg, 9.1–11.0 kg, 11.1–13.0 kg, and ≥13.1 kg. As the number of pre-pregnancy overweight or obese women was small, GWG in this group was classified as ≤5.0 kg and ≥5.1 kg. Among pre-pregnancy underweight and normal women, the incidence of SGA birth was higher than 10% in the categories of 9.0 kg or less GWG (Table 3). We set the subgroup of pre-pregnancy normal weight with 9.1–11.0 kg weight gain as a reference and calculated multiple-adjusted OR of the combined categories of pre-pregnancy BMI and GWG. Among pre-pregnancy underweight women, poor weight gain during pregnancy (9.0 kg or less) was associated with a remarkably increased incidence of SGA birth: multiple OR were 6.12 for ≤7.0 kg (95%CI 2.42–15.46, P < 0.001) and 7.57 for 7.1–9.0 kg (95%CI 3.09–18.58, P < 0.001) after adjustment for maternal age at entry, short stature, current smoking, drinking habit, and infertility treatment. On the other hand, pre-pregnancy underweight women who achieved 9.0 kg or more weight gain during pregnancy did not show a significantly greater risk of SGA birth. Among pre-pregnancy normal weight women, weight gain of 9.0 kg or less was associated with an increased risk of SGA birth. Pre-pregnancy overweight or obese women did not show any significantly increased risk for SGA in both subgroups according to GWG.

Table 3.  Multiple-adjusted OR for SGA birth according to the joint analyses of pre-pregnancy BMI and gestational weight gain
Pre-pregnancy BMI (kg/m2)Case/Total (%)Crude OR95%CI P Multiple-adjusted OR95%CI P
Gestational weight gain (kg)
  1. †SGA was defined as below the 10th percentile of birthweight at each gestational age, baby sex, and mother's parity. ‡Multiple-adjusted model was adjusted for maternal age at entry, short stature, smoking habit, drinking habit, and infertility treatment. BMI, body mass index; CI, confidence interval; OR, odds ratio; SGA, small-for-gestational-age.

Underweight (<18.5)       
 ∼7.09/33 (27.3)5.632.30–13.78<0.0016.122.42–15.46<0.001
 7.1∼9.010/33 (30.3)6.522.72–15.66<0.0017.573.09–18.58<0.001
 9.1∼11.05/72 (6.9)1.120.40–3.110.8281.350.48–3.790.573
 11.1∼13.03/42 (7.1)1.150.33–4.080.8241.240.35–4.470.740
  13.1 ∼2/30 (6.7)1.070.24–4.840.9291.180.25–5.500.835
Normal weight (18.5–24.9)       
 ∼7.023/196 (11.7)1.991.06–3.770.0332.111.10–4.030.024
 7.1∼9.030/251 (12.0)2.041.12–3.710.0202.211.19–4.080.012
 9.1∼11.019/304 (6.3)1.00 1.00 
 11.1∼13.018/192 (9.4)1.550.79–3.040.2001.520.77–3.010.232
  13.1 ∼6/143 (4.2)0.660.26–1.680.3810.630.24–1.640.346
Overweight and obese (≥25.0)       
 ∼5.05/33 (15.2)2.680.93–7.720.0682.540.86–7.510.092
  5.1 ∼162 (1.6)0.250.03–1.870.1760.220.03–1.710.149

We used additional multiple logistic regression models in order to examine the effect of pre-pregnancy BMI and GWG on delivery of infants below the 5th percentile of birthweight (Table 4). Greater GWG was associated with a decreased incidence of delivery of infants below the 5th percentile of birthweight. Pre-pregnancy underweight was associated with an increased risk of delivery of infants below the 5th percentile of birthweight. Each OR for delivery of infants below the 5th percentile of birthweight was similar to that for SGA birth defined as below the 10th percentile. We examined the incidence of delivery of infants below the 5th percentile among each pre-pregnancy BMI- and GWG-specific category. The cumulative incidence of delivery of infants below the 5th birthweight percentile was 12.1% in pre-pregnancy underweight women with 9.0 kg or less of GWG. This was much higher than that among pre-pregnancy normal women with 9.1–11.0 kg of GWG (2.0%). On the other hand, 4.2% of pre-pregnancy underweight women with more than 9.0 kg weight gain during pregnancy delivered infants below the 5th birthweight percentile. We did not perform further multivariate joint analysis of pre-pregnancy BMI and GWG in association with the risk of delivery of infants below the 5th percentile because of an insufficient number of events.

Table 4.  Multivariate model of the incidence of delivery of infants below the 5th percentile of birthweight
Independent variablesCrude OR95%CI P Multiple-adjusted OR95%CI P
  1. †We calculated the 5th percentile of birthweight at each gestational age, baby sex, and mother's parity. ‡Multiple-adjusted model was adjusted for all the variables listed here. BMI, body mass index; CI, confidence interval; OR, odds ratio.

Age at entry (years)      
 16–250.850.37–0.950.7000.810.34–1.900.624
 26–301.00  1.00  
 31–351.130.63–2.040.6861.130.62–2.060.701
 36–420.520.17–1.520.2300.540.18–1.620.268
Gestational weight gain (kg)0.880.81–0.960.0030.850.78–0.93<0.001
Pre-pregnancy BMI (kg/m2)      
 Underweight (<18.5)1.780.95–3.310.0712.111.11–4.000.023
 Normal weight (18.5–24.9)1.00  1.00  
 Overweight and obese (≥25.0)0.810.25–2.670.7300.390.11–1.400.149
Height (cm)      
 <1551.921.11–3.320.0201.931.10–3.400.022
 ≥1551.00  1.00  
Smoking habit      
 Non-current smoker1.00  1.00  
 Current smoker2.421.11–5.280.0263.811.64–8.830.002
Drinking habit      
 <1 time a week1.00  1.00  
 ≥1 time a week0.750.39–1.420.3710.640.32–1.250.188
Infertility treatment      
 No treatment1.00  1.00  
 Any treatment1.270.63–2.560.4971.400.67–2.890.370

Current smoking was strongly associated with the risk of SGA birth, while drinking habit did not show significant association with SGA birth (Table 2). Overall, infertility treatment was not associated with the incidence of SGA birth (Table 2). These tendencies were observed also regarding the risk of delivery of infants below the 5th percentile birthweight (Table 4).

Discussion

These prospective findings demonstrate that pre-pregnancy underweight and poor GWG were independently associated with an increased risk of SGA birth defined as below the 10th percentile of birthweight as well as delivery of infants below the 5th percentile of birthweight. Although pre-pregnancy underweight was an increased risk for SGA, pre-pregnancy underweight women who gained 9.0 kg or more during pregnancy did not show a significant association with greater incidence of SGA birth.

Only two previous studies from Japan have reported that low pre-pregnancy BMI and low GWG had detrimental effects on the incidence of SGA birth.3,4 Tsukamoto et al. reported in a historical study of 2972 Japanese women that lower pre-pregnancy BMI and lower GWG were associated with an increased risk of SGA birth.3 The multiple-adjusted OR for women with weight gain < 8.0 kg during pregnancy was 1.79 (95%CI 1.24–2.58) compared with those with weight gain of 10.1–12.0 kg. In their study, they did not examine the combined effects of both pre-pregnancy BMI and GWG upon SGA birth. Wataba et al. reported in a retrospective study of 21 718 women that lower pre-pregnancy BMI and poor GWG were associated with SGA. However, in their study, adjustment for covariates was lacking.4 The present study offers valuable insight into sufficient GWG for underweight women in Japan. We analyzed the impact of GWG according to pre-pregnancy BMI upon the incidence of SGA birth using joint analyses of pre-pregnancy BMI and GWG, which were not used in the other two Japanese studies. On the basis of the results of the present study that pre-pregnancy underweight women who gained more than 9.0 kg did not show significantly higher OR for the risk of SGA birth compared with the reference category, consisting of pre-pregnancy normal weight women with 9.1–11.0 kg of GWG, around 9.0 kg GWG or more might be a sufficient level for underweight women in Japan to reduce the risk of SGA birth. The Danish National Birth Cohort provided well-established data using a similar type of joint category11 and showed the sufficient level of GWG for underweight women: more than 15 kg for primiparous women and more than 20 kg for multiparous women.11 This discrepancy seems to be due to great differences in anthropometric factors between Danish and Japanese women. Given that over 30% of underweight women gained less than 9.0 kg during pregnancy in this study, we should focus on gaining sufficient GWG to prevent SGA birth rather than avoiding weight gain. The upper limit should be determined considering associations with other adverse outcomes, as we could not provide it with the current data.

We additionally examined the effects of pre-pregnancy BMI and GWG upon delivery of infants below the 5th percentile of birthweight. A multiple-adjusted logistic regression model revealed that both pre-pregnancy underweight and poor GWG were associated with an increased incidence of delivery of infants below the 5th percentile. Likewise, short stature was an increased risk for delivery of infants below the 5th percentile. All the OR were similar to those for SGA birth defined as below the 10th percentile of birthweight. On the other hand, the multiple-adjusted OR of current smoking for delivery of infants below the 5th percentile was greater than that for SGA birth defined as below the 10th percentile of birthweight: 3.81 for delivery of infants below the 5th percentile and 2.94 for SGA birth. Smoking might be a more critical factor as a risk for SGA birth than poor gestational weight gain and underweight.

The strength of our study was due to it being a single-center study. The authors could directly manage all the medical records of the participants, which could provide highly reliable data. Our study also has some possible limitations. First, our end-points depend on the percentile of our participants; therefore, the cut-off values would shift and results could be altered upon application to other populations. It is also uncertain whether our findings apply to other ethnic groups. The definition of SGA itself, that is, infants with birthweight less than the 10th percentile, had a limitation in that it could not exclude simply small infants who are not pathologically growth-restricted. Since no participants were malnourished, SGA infants born from pre-pregnancy underweight women and short women would be explained by a physiologically or constitutionally normal process. Second, information about maternal height and pre-pregnancy weight was self-reported, with the inevitable possibility of misclassification error. Third, smoking and drinking habit may change after answering the questionnaires in early pregnancy and such change could not be caught in the current study. Indeed, all the pregnant women, including the study participants in this hospital, were advised to quit smoking and drinking at prenatal visits and most of them were aware of the harm of smoking and alcohol. In addition, pregnant women tended to underreport their smoking habits.12 These misclassifications would cause only underestimation of the detrimental effects of smoking and drinking.

For Japanese and Asian pregnant women, there are no standard guidelines for GWG according to pre-pregnancy BMI. Lack of epidemiological data could make it difficult to obtain consensus on this issue. Application of the IOM recommendation does not seem practical. It was true that women with GWG within the recommendation range of the IOM guidelines did not have an increased risk of SGA birth in our study (data not shown). At the same time, however, more than 80% of pre-pregnancy underweight women in our study gained less weight than the recommendation. In addition, IOM defined short stature as less than 157 cm; almost half of our participants fitted this definition. Hence, accumulation of data from epidemiological studies is required to provide adequate guidelines for Japanese and other ethnic groups that are substantially shorter than the US population. This study suggested sufficient GWG for underweight women to prevent SGA birth using Japanese data, but further studies are required to assess adequate GWG for avoiding other adverse pregnancy outcomes.

Acknowledgments

This work was supported by Grant-in-Aid for Research Activity Start-up (22890163). The authors have no potential conflicts of interest to disclose. We thank Ms R. Higuchi and all the other medical clerks of Ishida Hospital for a great deal of assistance in managing the participants' data. Extending appreciation for the cooperation of all the study participants, we wish for the future good health of them and their children.

Disclosure

None declared.

Ancillary