Secondary Logo

Journal Logo

Original Article

Combined Effects of Prepregnancy Body Mass Index and Weight Gain During Pregnancy on the Risk of Preterm Delivery

Dietz, Patricia M.; Callaghan, William M.; Cogswell, Mary E.; Morrow, Brian; Ferre, Cynthia; Schieve, Laura A.

Author Information
doi: 10.1097/01.ede.0000198470.26932.9a


Click on the links below to access all the ArticlePlus for this article.

Please note that ArticlePlus files may launch a viewer application outside of your web browser.

Preterm delivery (before 37 weeks of gestation) persists in being one of the greatest contributors to infant mortality in the United States. Efforts to prevent this event have been relatively unsuccessful.1 Both low pregnancy weight gain and underweight prepregnancy body mass index (BMI; weight [kg]/height[m2]) are well-established risk factors for delivering before term.2,3 Recent studies have found that low pregnancy weight gain among underweight or normal-weight women increases preterm delivery risk, but the evidence regarding overweight and obese women has been equivocal. This lack of clear evidence may be due in part to different measures of weight gain, limited sample size, and different reference groups.4–7 Evidence for an effect of excessive weight gain on preterm delivery has also been equivocal. Although a few studies suggest such an association, they defined excessive weight gain differently, measured weight gain at different periods during pregnancy, inconsistently assessed interaction with BMI, and were restricted to low-income women.6–8

The objective of this study was to estimate the combined effects of prepregnancy BMI and pregnancy weight gain on preterm delivery of singleton births. Some risk factors for very preterm delivery differ from those for moderately preterm delivery, and so we examined these outcomes separately.


The large numbers of women available in the Pregnancy Risk Assessment Monitoring System (PRAMS) provided an opportunity to explore these associations in a population-based sample across the distribution of prepregnancy BMI and gestational weight gain. The Centers for Disease Control and Prevention (CDC) established PRAMS in 1987 to provide state-specific surveillance of maternal attitudes and experiences before, during, and shortly after pregnancy.9 Currently, 29 states and New York City conduct such surveillance. Our analysis used 1996–2001 PRAMS data from 21 states with annual response rates of 70% or higher: Alabama (1996–2001), Alaska (1996–2000), Arkansas (1997––2000), Colorado (1998–2001), Florida (1996–2001), Georgia (1996–1997), Hawaii (2000–2001), Illinois (1997–2001), Louisiana (1998–2000), Maine (1996–2001), Maryland (2001), Nebraska (2000–2001), New Mexico (1998–2000), New York (excluding New York City; 1996–2000), North Carolina (1997–2001), Ohio (1999–2000), Oklahoma (1996–2000), South Carolina (1996–2000), Utah (1999–2000), Washington (1996–2000), and West Virginia (1996–2000). The overall response rate was 78% (range, 70–81%).

Every month, in each state, a stratified, systematic sample of 100 to 200 new mothers was selected from birth certificates. Stratification variables were determined by the states, many of which included low birth weight. Each mother was mailed a 14-page questionnaire 1 to 3 months after delivery. If the mother failed to respond, additional questionnaires were mailed. If the mother did not respond to the questionnaires, attempts were made to interview her by telephone. The questionnaire was linked to the child's birth certificate. The final dataset included information from both the questionnaire and the birth certificate. PRAMS was approved by the CDC Institutional Review Board and women provided informed consent.

The data were weighted to adjust for survey design, noncoverage, and nonresponse. The weighted data are representative of all women who are state residents and who are delivering a live birth in the state, with the exception of New York. In that state, there are 2 reporting jurisdictions for vital records: New York City and New York State excluding the city. For New York State data in the year 2000, the final weights incorrectly weighted the data by approximately 4% because they included births to residents of New York City that were delivered outside of the city but within New York State.

We assigned gestational age for each birth by using a hierarchy that used 3 pieces of information from 2 different sources. The birth certificate provided information about last menstrual period (LMP) and clinical estimate of gestational age at birth, whereas the PRAMS questionnaire provided the mother's report of due date. LMP was considered the best source for assigning gestational age and was used if it was within 2 weeks of either the clinical estimate or the estimate based on the mother's reported due date (Appendix A). Otherwise, the clinical estimate was used if it was within 2 weeks of the mother's report of due date. For records still unassigned because of discrepancies or missing data, we estimated gestational age based on available date in the following order: LMP, clinical estimate, and PRAMS-based estimate. We also assessed gestational age assignment for plausibility with birth weight and excluded implausible estimates.10 The final gestational age estimate was categorized as 20–31 weeks (very preterm), 32–36 weeks (moderately preterm), and 37–44 weeks (term). We examined the birth weight distribution of each preterm group to assess misclassification and we assumed that a bimodal distribution would indicate misclassification. All birth weight distributions were unimodal.

Information used to calculate prepregnancy BMI was reported by the mother on the PRAMS questionnaire. Categories for prepregnancy BMI (kg/m2) were based on the Institute of Medicine (IOM) definitions: underweight, <19.8; normal, 19.8–26.0; overweight, 26.1–28.9; and obese, ≥29.0.2 We created an additional category for very obese (BMI ≥35) and redefined obese as 29.0–34.9.

Total pregnancy weight gain was recorded on the birth certificate. Previous studies have found that rate of weight gain is low in the first trimester and is higher and linear in the second and third trimesters, although slightly higher in the second trimester than in the third.2,11 If we assumed a constant rate of weight gain across trimesters, we would positively bias our results because women with shorter gestation would have a lower rate of weight gain. Accordingly, we estimated weight gain during the second and third trimesters by subtracting an assumed average weight gain during the first trimester. First-trimester weight gain was based on data in the IOM report: 2.3 kg (5 lbs) for underweight women, 1.6 kg (3.5 lbs) for normal-weight women, and 0.91 kg (2 lbs) for overweight, obese, or very obese women. We calculated a rate of weight gain during the second and third trimesters using the estimated weight gain for those 2 trimesters divided by the total gestational age (in weeks) minus 14 weeks. We divided the average weekly rate of weight gain into 5 groups to assess dose response: <0.12 kg/wk (very low, <0.25 lbs/wk), 0.12–.22 kg/wk (low, 0.25–0.49 lbs/wk); 0.23–0.68 kg/wk (moderate, 0.50–1.50 lbs/wk), 0.69–0.79 kg/wk (high, 1.51–1.75 lbs/wk), and >0.79 kg/wk (very high, >1.75 lbs/wk) (see Appendix B, available with the online version of this article). This definition of moderate weight gain included the weight gain recommended by the IOM if the pregnancy went to term: 12.7–18.1 kg (28–40 lbs) for underweight women, 11.3–15.9 kg (25–35 lbs) for normal-weight women, and 6.8–11.3 kg (15–25 lbs) for overweight or obese women. The other weight gain categories were created to ensure approximately equal sample sizes among them. To explore net weight gain, we recalculated the average weekly rate of weight gain after subtracting the infant birth weight from the total weight gain and then created the 5 weight gain category variables in the same manner as described.

Previous studies have shown that prepregnancy BMI modifies the association between weight gain during pregnancy and preterm delivery.4,6 We tested this interaction and confirmed it in our sample (P < 0.0001). We, therefore, constructed a 25-level variable that combined 5 levels of weight gain with 5 levels of prepregnancy BMI. Based on previous studies, we considered potential confounders and variables potentially in the causal pathway between weight gain and preterm delivery, including small for gestational age (SGA), diabetes (preexisting and gestational), and hypertension.1,2,12,13 Age, education, race, marital status, parity, infant birth weight, hypertension, and the trimester prenatal care was initiated were reported on the birth certificate. SGA was defined as birth weight less than the tenth percentile of birth weight distribution of a U.S. reference population.10 Maternal height, Medicaid recipient, pregnancy intention, and physical abuse by partner were reported on the PRAMS questionnaire. Information about diabetes (preexisting and gestational) and smoking during pregnancy was obtained from the birth certificate and the PRAMS questionnaire.

Our only exclusion criterion was multiple births. We examined associations of weight gain during pregnancy with very preterm and moderately preterm delivery by different levels of prepregnancy BMI. Multiple variable nominal polytomous logistic regression models were estimated for a 3-level outcome variable of very preterm, moderately preterm, and term delivery. The reference group for the independent variable of combined prepregnancy BMI and weight gain during pregnancy was women with normal prepregnancy BMI who gained an estimated 0.23–0.68 kg/wk in the second and third trimesters of pregnancy. Analyses were conducted for all women and for women without SGA, diabetes, or hypertension to assess whether associations remained in the absence of complications and conditions associated with medically indicated preterm delivery. Potential confounders were entered into the model individually and included in the final model if they changed associations by ≥10% (final model adjusted for maternal race–ethnicity, Medicaid recipient, parity, and marital status).14 SUDAAN (release 8.0; RTI International, Research Triangle Park, NC) was used for all analyses so that the standard errors reflected selection and response probabilities for the survey design. All analyses were conducted using weighted data.


Among respondents to the PRAMS questionnaire, 136,802 mothers with singleton births were eligible for the study. We excluded women with missing or implausible information on prepregnancy BMI (n = 11,518; 8.4%), on weight gain during pregnancy (n = 12,751; 9.3%), or on gestational age (n = 2113; 1.5%) (total exclusions: n = 23,783; 17.4% of the eligible sample). There was missing information for 24% of very preterm deliveries, 16% of moderately preterm deliveries, and 15% of term deliveries. Women with missing information were more likely to have <12 years of education (33% vs 18%), and be Medicaid recipients (61% vs 47%) and < 62 inches in height (41% vs 12%), but less likely to smoke during pregnancy (11% vs 14%). Women included and excluded from the sample were similar with respect to age, race, previous preterm birth, and delivering an SGA infant.

Overall, 8.7% of women delivered a preterm infant: 1.1% very preterm infant and 7.6% moderately preterm infant. Sixteen percent of women were considered underweight before pregnancy and 29% were overweight, obese, or very obese (Table 1). The majority of women gained between 0.69 and 0.79 kg/wk, were age 20–34 years, had 12 or more years of education, were married, and were white (Table 1).

Selected Characteristics Among Women With Moderately Preterm and Very Preterm Delivery: PRAMS 1996–2001

In general, as prepregnancy BMI increased, the amount of weight gained during pregnancy decreased (Table 2). For example, 2% of underweight women gained <0.12 kg/wk, whereas 19% of very obese women gained this amount.

Distribution of Weight Gain During Pregnancy by Prepregnancy Body Mass Index

Compared with women in the reference group, very low weight gain was associated with moderately preterm delivery among under-, normal-, and overweight women, and low weight gain was associated with this outcome among under- and normal-weight women (Table 3). These associations did not change after adjustments for maternal race–ethnicity, Medicaid recipient, parity, and marital status (data not shown), or after exclusions for diabetes, hypertension or SGA infants, and adjustments (Table 3). The associations between excessive weight gain (>0.79 kg/wk) and moderately preterm delivery were affected by exclusions (but not by adjustments for covariates) such that excessive weight gain increased the odds of moderately preterm delivery for under- and normal-weight women, but was no longer associated with moderately preterm delivery among heavier women. Similar results were found when we repeated the analysis using average weekly net weight gain (data not shown).

Odds for Moderately Preterm Delivery by Prepregnancy Body Mass Index and Weight Gain During Pregnancy

The associations between the combination of prepregnancy weight and weight gain and very preterm delivery were stronger than those for moderately preterm delivery (Table 4). Adjustments for covariates did not greatly affect the magnitude of the associations (data not shown), but exclusions for diabetes, hypertension, or SGA infants reduced the magnitude of the associations between excessive weight gain and very preterm delivery. After excluding women with diabetes, hypertension, or a SGA infant and adjusting for confounding variables, all BMI groups with very low weight gain had increased odds for very preterm delivery; however, as BMI increased, the odds decreased. Among women with very low weight gain, the odds of very preterm delivery were highest among women with underweight BMI (adjusted odds ratio [OR] = 9.8) and lowest among very obese women (adjusted OR = 2.3). Regardless of weight category, excessive gain (>0.79 kg/wk) substantially increased the risk of delivering very preterm (adjusted ORs = 2.0–2.8). Results were similar when we repeated the analysis using average net weight gain in kilograms per week (data not shown).

Odds for Very Preterm Delivery by Prepregnancy Body Mass Index and Weight Gain During Pregnancy


This is the first published study to examine the combined effect of prepregnancy BMI and weight gain during pregnancy for very preterm and moderately preterm delivery. Consistent with previous studies, we found that low weight gain among underweight and normal-weight women was associated with increased risk of preterm delivery.4–7 The magnitude of the association was greater for very preterm delivery than for moderately preterm delivery. In addition, very low weight gain (<0.12 kg/wk) was a risk factor for very preterm delivery among all BMI groups, including overweight and obese women; however, the magnitude of the odds ratios decreased as BMI increased. In addition, we found that excess weight gain (>0.79 kg/wk) among all BMI groups was associated with increased risk for very preterm delivery. Excess weight gain among under- and normal-weight women also increased the risk of moderately preterm delivery. When using net weight gain, our findings were similar to those for total weight gain.

Among women who gained very little during pregnancy, we found a generally declining risk for preterm delivery with increasing prepregnancy BMI. This suggests that having access to stored fat may protect against preterm delivery when weight gain is less than optimal. A similar association has been found for infant birth weight.15 Although weight restriction early in pregnancy has been associated with preterm delivery among sheep, it is unknown whether weight restriction later in pregnancy has the same effect, or whether this effect translates to humans.16 Some studies, however, suggest that fasting (which may be related to weight restriction) during late gestation may stimulate early preterm delivery.17,18 A review of 5 trials of supplementation with energy or protein during pregnancy among women at risk found only a modest reduction in preterm delivery. This suggests that simply increasing energy consumption during pregnancy does not lower the preterm delivery risk, perhaps because weight gain during pregnancy is multifaceted.19 In addition to representing energy stores, low weight gain may indicate deficiencies in micronutrients, a lack of expansion of plasma volume, infection, or other unidentified problems.3,20,21 The hypothesis that multiple mechanisms are at play is further supported by associations between low weight gain and both moderately and very preterm delivery. Moderately and very preterm deliveries are thought to represent some overlapping and some etiologically distinct outcomes: infection and inflammation are thought to be a greater underlying cause of very preterm delivery.1 The stronger association of low weight gain with very preterm delivery suggests that exploring the associations among prepregnancy weight, gestational weight gain, markers for infection or inflammation, and preterm delivery might be productive.

There are no clear biologic mechanisms for the link between excessive weight gain during pregnancy and preterm delivery. Other studies with adequate power have found results similar to ours.6–8 Excess weight gain may be associated with preterm delivery because it is a marker for edema, which, in turn, is a marker for preeclampsia.8 However, when we restricted our analysis to women without hypertension (presumably excluding most women with preeclampsia), the associations remained. One study found the association only with excessive third-trimester weight gain and only among women with normal prepregnancy BMI.7 In our study, after excluding women with hypertension, diabetes, or a SGA infant, we found an association with moderately preterm delivery for underweight and normal-weight women. Obesity and excess weight gain among postmenopausal women have been associated with at least one marker of inflammation, C-reactive protein,22,23 and a substantial percentage of preterm delivery is thought to be inflammatory in origin.24 It is unknown whether excess pregnancy weight gain is associated with inflammation. Our finding also may be an artifact of the lack of precision in our measurement of estimated weight gain in the second and third trimester. However, if real, our finding is relevant to a large proportion of pregnant women; in our population-based sample, more than one in 10 women gained >0.79 kg/wk during pregnancy.

This study benefited from a large sample size that allowed examination of preterm risks among overweight, obese, and very obese women separately, and assessment of risks for very and moderately preterm delivery. It was limited, however, by its measure of pregnancy weight gain. As reported on the birth certificate, total weight gain does not indicate weight gain by trimester. We assumed a constant rate of weight gain in the second and third trimesters, although studies have indicated that average third-trimester gain is slightly lower than average second-trimester gain.2,11 Our method of calculating estimated second- and third-trimester gain could not adjust for this difference. Thus, mothers of term infants might have slightly lower weight gain rates than mothers of very preterm infants due to longer exposure to lower rate of weight gain in the third trimester. A recent U.S. study found the mean weight gain among normal BMI women with term deliveries was 0.56 kg/wk in the second trimester and 0.50 kg/wk in the third trimester.11 Thus, for a woman who delivered at 40 weeks, the average weight gain in the second and third trimester would be 0.53 kg/wk, but for a woman who delivered at 26 weeks, the average weight gain would be 0.56 kg/wk. Although this bias may have contributed to the finding regarding excess weight gain, it is unlikely to have accounted for the entire association. Additionally, we assumed a set weight gain in the first trimester based on prepregnancy BMI as proposed by the IOM.2 We took this approach to account for potential bias of lower weight gain among women who deliver preterm because the lower rate of weight gain during the first trimester would have disproportionately contributed to the weight gain rate. We acknowledge there is likely misclassification of the weight gain rate, because some women will have gained more or less than what was assumed for the first trimester. We are unaware of any literature indicating this misclassification is differential among women delivering preterm and term infants.

Birth certificate data have other limitations. Problems with LMP-based gestational age reported on the birth certificate are well documented.25 We used 2 additional measures, clinical estimate and mother's report of due date, to check consistency with the LMP-based estimate. We also evaluated the birth weight distribution by gestational week and found no bimodal distribution, suggesting the misclassification with LMP-based gestational age was addressed by our gestational-age algorithm. Other birth certificate variables such as hypertension and diabetes are known to be underreported.26,27 Our final sample, in which we excluded complications, may have included preterm births related to these unreported conditions; thus, residual confounding may have remained. However, the magnitude of this bias, if present, would likely be insufficient to account for the entire association found in this analysis. Ideally, we would have liked to analyze these data by medically indicated and spontaneous preterm delivery, but PRAMS and the birth certificate lack this information. We tried to address this distinction by excluding women with the most common causes of medically indicated preterm delivery (hypertension, diabetes, and SGA). However, given the overlapping causes of both medically indicated and spontaneous preterm delivery, and the unknown etiology of the weight gain association with preterm delivery, it is unclear how including both types of preterm delivery might have affected our results. One study found that after medically indicated preterm births were excluded, the association between low prepregnancy BMI and low weight gain with preterm delivery was strengthened.4 We found a similar effect after our exclusions for missing information on weight gain during pregnancy, gestational age, or prepregnancy BMI. Variables in which the excluded women differed from those included were adjusted for in the logistic models, and this adjustment did not alter the observed associations. However, the generalizability of our study may be limited if the associations differ in the excluded women by unmeasured variables. Our preterm delivery rate is lower than the national rate because we excluded multiple births and our gestational age algorithm was a more restrictive criterion than using LMP alone.

In summary, very low weight gain was associated with moderately preterm delivery for the lowest BMI groups and with very preterm delivery among all BMI groups. In addition, although the strength of the association between excessive weight gain and preterm delivery was not as strong as the association for low weight gain, excessive weight gain affects a much greater proportion of the population and deserves further study. Further understanding of these associations is needed because it remains unclear whether they are causal and therefore amenable to nutritional interventions.


The following PRAMS Coordinators were responsible for coordinating data collection: Alabama, Rhonda Stephens; Alaska, Kathy Perham-Hester; Arkansas, Gina Redford; Colorado, Alyson Shupe; Florida, Helen Marshall; Georgia, Carol Hoban; Hawaii, Limin Song; Illinois, Theresa Sandidge; Louisiana, Joan Wightkin; Maine, Martha Henson; Maryland, Diana Cheng; Nebraska, Jennifer Severe-Oforah; New Mexico, Ssu Weng; New York State, Anne Radigan-Garcia; North Carolina, Paul Buescher; Ohio, Amy Davis; Oklahoma, Dick Lorenz; South Carolina, Sylvia Sievers; Utah, Lois Bloebaum; Washington, Linda Lohdefinck; and West Virginia, Melissa Baker.


1. Goldenberg RL, Rouse DJ. Prevention of premature birth. N Engl J Med. 1998;339:313–320.
2. Institute of Medicine. Nutrition During Pregnancy. Part I. Weight Gain. Washington, DC: National Academy Press; 1990.
3. Carmichael SL, Abrams B. A critical review of the relationship between gestational weight gain and preterm delivery. Obstet Gynecol. 1997;89:865–873.
4. Schieve LA, Cogswell ME, Scanlon KS, et al. Prepregnancy body mass index and pregnancy weight gain: associations with preterm delivery. The NMIHS Collaborative Study Group. Obstet Gynecol. 2000;96:194–200.
5. Spinillo A, Capuzzo E, Piazzi G, et al. Risk for spontaneous preterm delivery by combined body mass index and gestational weight gain patterns. Acta Obstet Gynecol Scand. 1998;77:32–36.
6. Schieve LA, Cogswell ME, Scanlon KS. Maternal weight gain and preterm delivery: differential effects by body mass index. Epidemiology. 1999;10:141–147.
7. Siega-Riz AM, Adair LS, Hobel CJ. Institute of Medicine maternal weight gain recommendations and pregnancy outcome in a predominantly Hispanic population. Obstet Gynecol. 1994;84:565–573.
8. Wen SW, Goldenberg RL, Cutter GR, et al. Intrauterine growth retardation and preterm delivery: prenatal risk factors in an indigent population. Am J Obstet Gynecol. 1990;162:213–218.
9. Adams MM, Shulman HB, Bruce C, et al. The pregnancy risk assessment monitoring system: design, questionnaire, data collection and response rates. Paediatr Perinat Epidemiol. 1991;5:333–346.
10. Alexander GR, Himes JH, Kaufman RB, et al. A United States national reference for fetal growth. Obstet Gynecol. 1996;87:163–168.
11. Carmichael S, Abrams B, Selvin S. The pattern of maternal weight gain in women with good pregnancy outcomes. Am J Public Health. 1997;87:1984–1988.
12. Goldenberg RL, Iams JD, Mercer BM, et al. What we have learned about the predictors of preterm births. Semin Perinatol. 2003;27:185–193.
13. Abrams B, Carmichael S, Selvin S. Factors associated with the pattern of maternal weight gain during pregnancy. Obstet Gynecol. 1995;86:170–176.
14. Maldonado G, Greenland S. Simulation study of confounder-selection strategies. Am J Epidemiol. 1993;130:923–936.
15. Abrams BF, Laros RK. Prepregnancy weight, weight gain, and birth weight. Am J Obstet Gynecol. 1986;154:503–509.
16. Bloomfield FH, Oliver MH, Hawkins P, et al. A periconceptional nutritional origin for noninfectious preterm birth. Science. 2003;300:600.
17. Siega-Riz AM, Herrmann RS, Savitz DA, et al. Frequency of eating during pregnancy and its effect on preterm delivery. Am J Epidemiol. 2001;153:647–652.
18. Herrmann TS, Siega-Riz AM, Hobel CJ, et al. Prolonged periods without food intake during pregnancy increase risk for elevated maternal corticotrophin-releasing hormone concentrations. Am J Obstet Gynecol. 2001;185:403–412.
19. Kramer MS, Kakuma R. Energy and protein intake in pregnancy (Cochrane Review). In: The Cochrane Library, Issue 2. Chichester, UK: John Wiley & Sons, Ltd; 2004.
20. Romero R, Chaiworapongsa, Espinoza J. Micronutrients and intrauterine infection, preterm birth and the fetal inflammatory response syndrome. Am Society Nutr Sciences. 2003:1668S–1673S.
21. Siega-Riz AM, Promislow JHE, Savitz DA, et al. Vitamin C intake and the risk of preterm delivery. 2003;189:519–525.
22. Barinas-Mitchell E, Cushman M, Meilahn EN, et al. Serum levels of C-reactive protein are associated with obesity, weight gain, and hormone replacement therapy in healthy postmenopausal women. Am J Epidemiol. 2001;153:1094–1101.
23. Saito I, Yonemasu K, Inami F. Association of body mass index, body fat, and weight gain with inflammation markers among rural residents in Japan. Circ J. 2003;67:323–329.
24. Goldenberg RL, Hauth JC, Andrews WW. Mechanisms of disease: intrauterine infection and preterm delivery. N Engl J Med. 2000;342:1500–1507.
25. Parker JD, Schoendorf KC. Implications of cleaning gestational age data. Paediatr Perinat Epidemiol. 2002;16:181–187.
26. DiGiuseppe DL, Aron DC, Random L, et al. Reliability of birth certificate data: a multi-hospital comparison to medical records information. Matern Child Health J. 2002;6:169–179.
27. Dobie SA, Baldwin L, Rosenblatt RA, et al. How well do birth certificates describe the pregnancies they report? The Washington State experience with low-risk pregnancies. Matern Child Health J. 1998;2:145–154.


Hierarchy for Assigning Gestational Age

Supplemental Digital Content

© 2006 Lippincott Williams & Wilkins, Inc.