Objective
The purpose of this study was to estimate the impact of interpregnancy weight change from first to second pregnancies in obese women on the risk of large-for-gestational-age (LGA) and small-for-gestational-age (SGA) infants.
Study Design
A population-based historical cohort analysis of 10,444 obese women in Missouri who delivered their first 2 singleton live infants from 1998-2005. Interpregnancy weight change was calculated as the difference between prepregnancy body mass index (BMI) of the first and second pregnancies. LGA and SGA births were compared among 3 interpregnancy weight change groups: (1) weight loss (≥2 BMI units), (2) weight gain (≥2 BMI units), and (3) reference group (BMI maintained within 2 units). Adjusted odds ratios (aOR) were calculated for LGA and SGA births with the use of multiple logistic regression. A dose-response relationship was assessed with a linear-by-linear χ 2 test.
Results
Compared with the reference group, interpregnancy weight loss was associated with lower risk of an LGA infant (aOR, 0.61; 95% confidence interval, 0.52–0.73), whereas interpregnancy weight gain was associated with increased risk of an LGA infant (aOR, 1.37; 95% confidence interval, 1.21–1.54). Interpregnancy BMI change was not related to SGA infant risk, except for weight loss of >8 BMI units . A significant dose-response relationship was observed for LGA infant risk ( P < .001), but not SGA infant risk ( P = .840).
Conclusion
Mild-to-moderate interpregnancy weight loss in obese women reduced the risk of subsequent birth of LGA infants without increasing the risk of SGA infants. The interpregnancy interval may be a crucial period for targeting weight loss in obese women.
Obesity has reached epidemic proportions in the United States, increasing dramatically over the past 20 years. Current obesity trends project that, by the year 2030, 51% of US adults will be obese. More than one-quarter of US women currently begin their pregnancies as obese. Furthermore, the severity of obesity is also increasing; the prevalence of class III obesity (body mass index [BMI], ≥40 kg/m 2 ) in women of reproductive age has tripled over the past 30 years.
Obesity is a modifiable condition that is associated with significant health consequences for both mothers and offspring. Obese women tend to retain more weight after delivery, which leads to a further increase in obesity and its comorbidities later in life. Prepregnancy obesity contributes to adverse neonatal and maternal outcomes, which include macrosomia, large-for-gestational-age (LGA) infants, composite neonatal morbidity, pregnancy-induced hypertension, gestational diabetes mellitus, and cesarean delivery. Furthermore, increasing the severity of obesity significantly elevates the risk of these adverse outcomes.
Prepregnancy obesity has a dose-dependent positive relationship with the risk of macrosomia and LGA infant delivery. LGA infants have greater adiposity at birth and an increased risk of obesity and metabolic syndrome during childhood and adolescence. In contrast, obesity has an inverse association with the risk of a small-for-gestational-age (SGA) infant. SGA infants are predisposed to a number of long-term health sequelae, which includes an increased risk of metabolic syndrome later in life.
Given the increasing prevalence of obesity among women of reproductive age, which places them at increased risk for both postpartum weight retention and LGA births, it is important to study the impact of interpregnancy weight change on birthweight of the offspring in obese women. Two previous studies have demonstrated an association between interpregnancy BMI change and LGA risk in the subsequent pregnancy in overweight and obese women. However, the effect of weight loss between pregnancies in obese women requires additional focus. Interpregnancy weight loss may offer the potential to improve both maternal and neonatal outcomes for obese women in the subsequent pregnancy. The aim of this population-based study was to determine (1) the impact of interpregnancy BMI change in obese pregnant women on the risk of delivering an LGA or SGA infant in the subsequent pregnancy and (2) the dose-dependent relationship between changes in maternal interpregnancy BMI and the risk of an LGA or SGA infant.
Materials and Methods
This population-based, historical, cohort study was conducted with data from the Missouri maternally linked birth and fetal death registry, which links maternal health data longitudinally to infant birth certificate data with the use of unique identifiers. The methods and algorithm that are used within the Missouri vital record system have been described previously, are considered reliable, and have been used to validate US national datasets that involve matching and linking procedures.
There were 69,555 eligible women included within the registry during the study years 1998-2005. Eligible women resided in Missouri and delivered their first 2 singleton live births without congenital malformations at 20-42 weeks’ gestation. Included in this analysis were 10,672 women who started their first pregnancy as obese (BMI, ≥30 kg/m 2 ), which was 15.3% of the total study population. Prepregnancy weight of the second pregnancy was missing for 228 women (2.1%). Therefore, the final cohort for analysis consisted of 10,444 women.
Maternal prepregnancy BMI was calculated from self-reported weight and height records from the Missouri dataset. BMI was calculated as weight (in kilograms) divided by the squared height (in meters). Participants were placed into 3 study groups on the basis of interpregnancy weight change, which was defined as a change in prepregnancy BMI from the first to second pregnancy. We deemed a change of at least 2 BMI units as a realistic, but clinically relevant, weight change. On the basis of this definition, our 3 study groups included (1) women who decreased their BMI by ≥2 units (weight loss group), (2) women who increased their BMI by ≥2 units (weight gain group), and (3) women who maintained their BMI within a 2 unit loss or gain (reference group).
The primary outcome variables for this study were LGA and SGA infants, defined as birthweight >90th percentile and <10th percentile, respectively, that were corrected for sex and gestational age. Potential demographic, obstetric, and medical confounders that could impact the risk of LGA and SGA infants in the second pregnancy were collected and adjusted for in the study. Demographic variables included maternal age (18-35 years, >35 years), race and ethnicity (non-Hispanic white, non-Hispanic African American, other), marital status (married, unmarried), education (≤12 years, >12 years), and socioeconomic status (using Medicaid enrollment as a proxy). The obstetric and medical confounders included baseline obesity status that was indicated by prepregnancy BMI of the first pregnancy (categorized as class I [30-34.9 kg/m 2 ], class II [35-39.9 kg/m 2 ] or class III [≥40 kg/m 2 ]), interpregnancy interval (<12 months, 12-36 months, and >36 months), and the following characteristics of the second pregnancy: gestational weight gain (GWG; in pounds), gestational age (in weeks), smoking, preeclampsia, adequacy of prenatal care, birth of an LGA or SGA infant in the first pregnancy, diabetes mellitus, chronic hypertension, renal disease, and cardiac disease.
Interpregnancy interval was calculated as the time (in months) between first and second births excluding clinical gestational age of the second pregnancy. Gestational age was determined by clinical estimate (a required field on the birth certificate since 1989), which was estimated by early second-trimester ultrasound scans. This has been found to be a more accurate measure of gestational age at delivery than determination by the last menstrual period. Adequacy of prenatal care was determined with the use of the Kotelchuck index, which is an algorithm that compares the observed number of prenatal care visits to the expected number of prenatal care visits based on the gestational age at initiation of care . Prenatal care was grouped into 4 categories: inadequate (received <50% of expected visits), intermediate (50-79%), adequate (80-109%), and adequate plus (≥110%). Diabetes mellitus was defined on the birth certificate as insulin-dependent diabetes mellitus or other diabetes mellitus, which did not distinguish between preexisting and gestational diabetes mellitus. Therefore, all women with diabetes mellitus, regardless of preexisting or gestational, were combined into a single group.
Demographic, medical, and obstetric characteristics in the second pregnancy were compared across interpregnancy BMI change groups with the χ 2 test for categoric variables, analysis of variance for continuous variables that were normally distributed, and the nonparametric Kruskal-Wallis test for continuous variables that were not normally distributed. The multivariate predictability of interpregnancy BMI change groups for an LGA infant was examined by multiple logistic regression analysis.
Multivariate findings reflect adjustment for all potential confounders in the model. To assess whether the severity of obesity had an impact on the relationship between interpregnancy BMI change and LGA infant risk, the interaction term between obesity class at the start of their first pregnancy and interpregnancy BMI change was included in the regression model. For dose-response analyses, interpregnancy weight loss and weight gain groups were categorized in 2 BMI unit increments. The incidence of LGA by interpregnancy BMI change was assessed by linear-by-linear χ 2 test for trend. Parallel analyses were performed for SGA infants. A probability value of < .05 was used to denote statistical significance. All analyses were performed with SPSS software (version 18.0 for Windows [Microsoft Corporation, Redmond, WA]; SPSS Inc, Chicago, IL).
The Saint Louis University Institutional Review Board deemed that the study did not require formal review under exemption 45 CFR 46.102, because the study did not involve interaction with individuals nor did the dataset contain identifiable private information.
Results
The study population of 10,444 obese women consisted of 1739 women (16.6%) who lost ≥2 BMI units between pregnancies, 4743 women (45.4%) who maintained their BMI between pregnancies (gain or loss, <2 BMI units), and 3962 women (38.9%) who gained ≥2 BMI units between pregnancies. In the entire study population, the prevalence of class II obesity rose from 25.5% in the first pregnancy to 33.2% in the second pregnancy. The prevalence of class III obesity rose from 15.2-27.3%.
Changes in BMI across first and second pregnancies are displayed in Figure 1 . Compared with the reference group, the weight gain group had higher GWG in their first pregnancy (median of 32 pounds [interquartile range (IQR), 22–42 pounds] vs reference group 25 pounds [IQR, 16–34 pounds]; P < .001) and a longer interpregnancy interval (23.75 months [IQR, 14–37 months] vs 19.25 months [IQR, 12–29 months]; P < .001). In contrast, women in the weight loss group had a higher BMI at the start of their first pregnancy (34.5 kg/m 2 [IQR, 30.2–38.8 kg/m 2 ] vs reference group 33.7 kg/m 2 [IQR, 31.6–37.3 kg/m 2 ]; P < .001) but lost weight between pregnancies, which resulted in a significantly lower BMI at the start of their second pregnancy (29.8 kg/m 2 [IQR, 27.4–33.7 kg/m 2 ] vs reference group 34.0 kg/m 2 [IQR, 31.8–37.6 kg/m 2 ] and weight gain group 39.0 kg/m 2 [IQR, 36.1–42.8 kg/m 2 ]; P < .001).
Demographic, obstetric, and medical characteristics of the second pregnancy across interpregnancy BMI change groups are displayed in Table 1 . Compared with the reference and weight gain groups, women in the weight loss group were more likely to be non-Hispanic African American, unmarried, classified at baseline as class III obese, have higher GWG in their second pregnancy, and be a smoker. Interpregnancy weight change had a significant effect on both diabetes mellitus and preeclampsia. Compared with the reference group, the diagnoses of diabetes mellitus and preeclampsia were increased in the weight gain group and decreased in the weight loss group. Interpregnancy weight change was not associated with alterations in gestational age at delivery, the risk for cardiac disease or chronic hypertension, a history of LGA or SGA infant, or the adequacy of prenatal care.
| Variable | Interpregnancy change in body mass index | ||
|---|---|---|---|
| Weight loss a (n = 1739) | Reference b (n = 4743) | Weight gain c (n = 3962) | |
| Demographic, n (%) | |||
| Maternal age d | |||
| 18-35 y | 1705 (98.0) | 4608 (97.2) | 3882 (98.0) |
| >35 y | 34 (2.0) | 135 (2.8) | 80 (2.0) |
| Maternal race and ethnicity d | |||
| Non-Hispanic white | 1476 (84.9) | 4136 (87.2) | 3432 (86.6) |
| Non-Hispanic African American | 234 (13.5) | 553 (11.7) | 492 (12.4) |
| Other | 22 (1.3) | 41 (0.9) | 27 (0.7) |
| Marital status e | |||
| Married | 1248 (71.8) | 3826 (80.7) | 2971 (75.0) |
| Unmarried | 491 (28.2) | 917 (19.3) | 991 (25.0) |
| Education e | |||
| ≤12 y | 558 (32.1) | 1442 (30.4) | 1424 (36.9) |
| >12 y | 1172 (67.4) | 3285 (69.3) | 2527 (63.8) |
| Medicaid recipient e | 689 (39.8) | 1640 (34.7) | 1730 (43.7) |
| Obstetric | |||
| Obesity status, first pregnancy, n (%) e | |||
| Class I obese (30–34.9 kg/m 2 ) | 923 (53.1) | 2868 (60.5) | 2403 (60.6) |
| Class II obese (35–39.9 kg/m 2 ) | 459 (26.4) | 1194 (25.2) | 1013 (25.6) |
| Class III obese (≥40 kg/m 2 ) | 357 (20.5) | 681 (14.3) | 546 (13.8) |
| Interpregnancy interval, n (%) e | |||
| <12 mo | 458 (26.3) | 1410 (29.7) | 865 (21.8) |
| 12-36 mo | 975 (56.1) | 2723 (57.4) | 2220 (56.0) |
| >36 mo | 284 (16.3) | 603 (12.7) | 874 (22.1) |
| Gestational weight gain, lb e , f | 31.0 (21-43) | 24.0 (15-35) | 19.0 (9-28) |
| Gestational age, wk f | 39.0 (38-40) | 39.0 (38-40) | 39.0 (38-40) |
| Smoked during pregnancy, n (%) e | 349 (20.1) | 695 (14.7) | 580 (14.6) |
| Preeclampsia, n (%) d | 91 (5.2) | 288 (6.1) | 281 (7.1) |
| Prenatal care, n (%) | |||
| Inadequate | 96 (5.5) | 231 (4.9) | 174 (4.4) |
| Intermediate | 164 (9.4) | 461 (9.7) | 390 (9.8) |
| Adequate | 778 (44.7) | 2175 (45.9) | 1788 (45.1) |
| Adequate plus | 641 (36.9) | 1740 (36.7) | 1522 (38.4) |
| Previous large-for-gestational-age infant, n (%) | 240 (13.8) | 635 (13.4) | 510 (12.9) |
| Previous small-for-gestational-age infant, n (%) | 171 (9.8) | 427 (9.0) | 322 (8.1) |
| Medical, n (%) | |||
| Diabetes mellitus: insulin-dependent and other d | 129 (7.4) | 379 (8.0) | 362 (9.1) |
| Chronic hypertension | 49 (2.8) | 161 (3.4) | 133 (3.4) |
| Renal disease e | 13 (0.7) | 10 (0.2) | 10 (0.3) |
| Cardiac disease | 9 (0.5) | 27 (0.6) | 31 (0.8) |
a Weight loss group: maternal race (n = 7); education (n = 9); Medicaid (n = 6); smoking (n = 3); weight gain in second pregnancy (n = 16); interpregnancy interval (n = 22); prenatal care (n = 60);
b Reference group: maternal race (n = 13); education (n = 16); Medicaid (n = 10); smoking (n = 5); weight gain in second pregnancy (n = 44); interpregnancy interval (n = 7); prenatal care (n = 136);
c Weight gain group: maternal race (n = 11); education (n = 11); Medicaid (n = 6); smoking (n = 1); weight gain in second pregnancy (n = 52); interpregnancy interval (n = 3); prenatal care (n = 88);
The overall incidence of LGA and SGA infants in the second pregnancy was 18.1% and 5.2%, respectively. Across all study groups, the incidence of LGA increased and SGA decreased in the second, compared with the first, pregnancy. Furthermore, the incidence of LGA was much higher than the incidence of SGA for all study groups. Infant birthweight increased with parity (median birthweight of whole cohort, 3415 g vs 3487 g). Despite this increase, second pregnancy offspring of women in the weight loss group were on average 56 g lighter compared with the reference group (median, 3459 g [IQR, 3124–3771 g] vs 3515 g [IQR, 3175–3827 g]). In contrast, infant birthweight was not affected by interpregnancy weight gain (3515 g [IQR, 3175–3853 g]).
The incidence and multivariate findings for LGA infants by interpregnancy weight change are given in Table 2 . Baseline obesity status did not modify the association between interpregnancy weight change and infant outcomes (interaction effect, P > .05 for the multiple logistic regression). Compared with the reference group, women in the weight loss group had significantly lower odds of delivering an LGA infant, whereas women in the weight gain group had significantly higher odds of delivering an LGA infant. In contrast, interpregnancy weight change was not significantly related to SGA infant risk.
