- Email: [email protected]
Macrosomia: Influence of maternal overweight among a low-income population
lams et al.
tus and uterine activity. At present it seems prudent to educate all pregnant women about these prodromal preterm labor symptoms, much as is done for the equally nonspecific common early symptoms of preeclampsia. REFERENCES
February 1990 Am J Obstet Gynecol
2. Creasy RK. Preterm labor and delivery. In: Creasy RK, Resnik R, eds. Maternal-fetal medicine principles and practice. Philadelphia: WB Saunders, 1984:415. 3. March of Dimes Birth Defects Foundation. Prevention of preterm labor and low birth weight package. White Plains, New York: March of Dimes, 1983. 4. Newman RB, Gill PJ, Wittreich P, Katz M. Maternal perception of prelabor uterine activity. Obstet Gynecol 1986; 68:765-9.
1. Herron MA, Katz M, Creasy RK. Evaluation of a preterm birth prevention program: preliminary report. Obstet Gynecol 1982;59:452-6.
Macrosomia: Influence of maternal overweight among a low-income population Christine E. Larsen, MD, MS, Mary K. Serdula, MD, MPH, and Kevin M. Sullivan, MPH Atlanta, Georgia Macrosomia (birth weight >4000 gm) is associated with a wide variety of adverse intrapartum and perinatal outcomes. To evaluate the effect of pregravid obesity on infant birth weight, we examined data from a low-income population of women (n = 127,879). The population was divided into five groups on the basis of pregravid body mass index (weight/heighf') designated by the National Health and Nutrition Examination Survey II reference population «25th percentile, 25th to <75th percentile, 75th to <85th percentile, 85th to <95th percentile, and 2!95th percentile). The prevalence of infant macrosomia ranged from 5% for the lowest group to 17% for the highest group. With the use of the second group (25th to <75th percentile) as a reference, odds ratios (adjusted for maternal age, smoking status, race, height, parity, gestational age, and infant sex) for macrosomia for the five sequential weight groups were 0.6, 1.0, 1.3, 1.6 and 2.2. We conclude that pregravid overweight had a significant independent effect on birth weight outcome. (AM J OSSTET GVNECOL 1990;162:490-4.)
Key words: Fetal macrosomia, birth weight, obesity
Macrosomia (birth weight >4000 gm) is associated with increased risk of infant birth trauma, fetal death, and cesarean deliveries. l Early recognition of maternal factors that contribute to macrosomia may decrease the incidence of birth injuries. 2 Because the prevalence of macrosomia increased in the U.S. population from 8.2% in 1965' to 11.1 % in 1984: the potential morbidity from macrosomia may escalate. Obese gravidous women have higher levels of hypertension, hyperglycemia, prolonged labor postamniotomy, postpartum hemorrhage, and puerperal pyrexia than do pregnant women who are not obese." From the Division of Nutrition, Center for Health PromotIOn and Education, Centers for DISease Control. Received for publication November 28, 1988; revised May 26, 1989; accepted August 7, 1989. Reprint requests: Mary K. Serdula, MD, MPH, DzvlSlon of Nutrition, Centers for DISease Control, Mazlstop A-41, Atlanta, GA
30333. 611/15881
490
Because the prevalence of macrosomia is higher among infants born to obese mothers than among infants born to nonobese mothers,6 macrosomia is of particular concern in low-income populations in which obesity is a major health problem. 7 Furthermore, because lowincome populations may have limited access to medical care, they may incur even greater risks associated with macrosomia. Although the association between maternal weight and macrosomia has been well documented, the observed association may have been a result of factors other than maternal overweight per se. Previous studies that examined the association between maternal obesity and mauosomia have not adequately accounted for other variables related to macrosomia. In addition, only limited research is available on the prevalence and risk factors for macrosomia among low-income populations. Our study evaluated the influence of pregravid
Influence of maternal overweight 491
Volume 162 Number 2
Table I. Characteristics of study population by ethnic group (%) WhIte
n
= 64,677
Black
n
Body mass index (NHANES II percentiles) 28.0 :S:25th 25th to <75th 43.3 75th to <85th 9.5 11.4 85th to <95th 2:95th 7.8 Maternal height (quartiles) 15.9 First 27.3 Second Third 29.1 27.8 Fourth 40.9 Smoking (yes) Maternal age (yr) 39.1 15-20 29.9 20-25 23.1 25-30 8.0 2:30 Previous live births 43.8 0 30.5 I 16.1 2 2:3 9.7 Gestational age (wk) 7.8 37 14.3 38 22.6 39 26.2 40 18.9 41 10.2 42
weight on birth weight outcome among a low-income population attending the Special Supplemental Food Programs for Women, Infants, and Children (WIC). We analyzed this influence after adjustment for the effects of smoking status, ethnicity, maternal height, maternal age, gestational age, and parity. Material and methods
Through its Pregnancy Nutrition Surveillance System, the Division of Nutrition of the Centers for Disease Control receives information from women who attend the WIC program. WIC records consist of maternal and infant information obtained at prenatal visits and at a postpartum visit. For this study we analyzed data from term (37 through 42 weeks' gestation), singleton live births born between 1983 and 1987. Infants weighing < 1000 gm or >6000 gm were excluded from the data analysis to minimize the likelihood of coding errors. Any records with missing or out-of-range values for variables used in this analysis were excluded. Only records from the white, black, and Hispanic ethnic groups were included. These exclusions left 127,879 records from 14 states. Data items were recorded by WIC clinic personnel. All data entries comprised information provided by the WIC client. No information was available on the prevalence of gestational diabetes, hypertension, and fetal anomalies. We examined the effects of pre gravid body mass in-
= 47,237
n
Hispanzc
= 15,965
n
=
Total
127,879
20.7 44.5 11.8 13.8 9.3
17.4 47.7 14.0 14.6 6.2
24.0 44.3 10.9 12.7 8.2
16.4 26.4 28.1 29.1 20.8
44.5 30.6 17.0 8.0 1l.8
19.7 27.4 27.2 25.8 29.9
36.3 27.8 25.2 10.7
26.0 27.9 28.9 17.3
36.4 28.8 24.6 10.2
33.0 32.1 19.0 15.9
25.2 35.2 21.6 18.1
37.5 31.7 17.9 13.0
Il.l
8.5 16.6 24.8 26.6 15.5 7.9
9.1 16.0 23.7 25.8 16.6 8.8
18.0 24.9 24.9 13.9 7.3
dex (BMI) on both infant macrosomia and mean infant birth weight. The BMI was the independent variable used in this study and was calculated by dividing selfreported prepregnant weight (kilograms) by height 2 (meters squared). The population was grouped into five categories on the basis of the percentile distribution of BMIs from the National Health and Nutrition Examination Survey (NHANES) II reference sample for women ages 20 to 29 years: <25th percentile «20.0 BMI units), 25th to <75th percentile (2:20.0 to 24.9 BMI units), 75th to <85th percentile (2:24.9 to 27.3 BMI units), 85th to <95th percentile (2:27.3 to 32.3 BMI units), and 2:95th percentile (2:32.3 BMI units)." The reference group used for statistical comparison was the 25th to <75th percentile group from NHANES II. This group most closely approximates an ideal body weight category. Women with BMIs in the 85th to <95th percentile were categorized as overweight, and those in the 2:95th percentile were categorized as severelyoverweight. Maternal height was divided into the following approximate quartiles: first (:s: 157.0 cm), second (157.1 to 162.2 cm), third (162.3 to 167.3 cm), and fourth (2: 167.4 cm). Smoking status (yes, no) was reported by the mother on her first prenatal WIC visit. Infant birth weight was reported by the mother at the postpartum visit. Gestational age was calculated on a weekly basis from the self-reported estimated date of confinement and the infants' birthdate. If the estimated
492 larsen, Serdula, and Sullivan
February 1990 Am J Obstet Gynecol
Table II. Prevalence of macrosomia and odds ratios for selected factors Macrosomic infants (%)
Adjusted odds ratio (95% Cl)*
Body mass index (NHANES II percentile) s25th 4.8 0.58 (0.55-0.62) 25th to <75th 8.1 LOOt 75th to <85th 10.8 1.34 ( 1.26-1.43) 85th to <95th 12.8 1.61 (1.52-1.70) ~95th 17.0 2.15 (2 .01-2.29) Maternal height (quartiles) First 5.9 1.00t Second 7.4 1.41 ( 1.32-1.52) Third (1.81-2.07) 9.3 1.93 Fourth 12.5 2.81 (2.63-3.01) Ethnic group White 11.1 LOOt Black 5.6 0.38 (0.36-0.40) Hispanic 10.1 0.86 (0.81-0.92) Smoking No 10.5 LOOt Yes 5.2 0.40 (0.38-0.42) Maternal age (yr) 15-19 6.9 (0.88-0.99) 0.93 20-24 8.7 LOOt 25-29 10.3 1.09 (1.03-1.15) ~30 13.5 1.33 ( 1.25-1.42) Previous live births 0 7.3 LOOt I 8.8 (1.18-1.32) 1.25 10.3 2 1.41 ( 1.32-1.50) ~3 12.0 1.54 (1.43-1.65) Gestational age (wk) 37 3.9 0.38 (0.34-0.42) 38 5.1 0.49 (0.46-0.53) 39 7.1 0.69 (0.65-0.73) 40 10.1 1.00t 41 13.2 1.34 (1.26-1.41 ) 42 14.7 1.55 (1.45-1.66) Infant sex Male 11.2 LOt Female 6.7 0.52 (0.50-0.54) *Multiple logistic regression with BMI, height, ethnicity, smoking, maternal age, previous live births, gestational age, and infant sex in the model. tReference group. date of confinement was not available, then gestational age was calculated from the maternally reported last menstrual period. In the univariate analysis, we computed the prevalence of macrosomia by BMI category. Our multiple logistic regression models used macrosomia as the dependent variable to adjust for the mother's BMI (five categories), height (quartiles), smoking status (yes, no), previous live births (0, 1, 2, ~3), maternal age (15 to 19, 20 to 24, 25 to 29, ~30 years), ethnicity (white, black, Hispanic), gestational age (37, 38, 39, 40, 41 , 42 weeks), and infants' sex. The magnitude and direction of associations between macrosomia and the independent variables were estimated by the odds ratio. The effects of maternal BMI on mean birth weight were quantified with multiple linear regression 'o according Q
to the same procedures as described for the logistic model. Results
Of the sample population, 51 % were white, 37% were black, and 12% were Hispanic (Table I). The prevalence of smoking varied from 41 % for whites to 21 % for blacks and 12% for Hispanics. The prevalences for mothers with BMIs ~85th percentile of the NHANES II reference population were 19% for whites, 23% for blacks, and 21 % for Hispanics. As the maternal BMI increased, the prevalence of macrosomia also increased (Table II). The prevalence of macrosomia ranged from 5% for those <25th percentile to 17% for those in the ~95th percentile. After adjustment for maternal height, ethnicity, smoking status, maternal age, parity, gestational age, and infant sex we also found a statistically significant increase in the odds ratios when the BMIs increased. Specifically, with the NHANES II category for the 25th to <75th percentile as a reference group, we found that the adjusted odds ratios of macrosomia for the five sequential weight groups were 0.6, 1.0, 1.3, 1.6 and 2.2 (Table II). In Table II we also present the adjusted odds ratios for macrosomia according to other variables such as maternal height, ethnicity, maternal age, smoking status, parity, gestational age, and infant sex. Maternal height appeared to have a strong influence : the tallest women were three times more likely to have macrosomic infants than were the shortest women. Mothers with a late gestational age of 42 weeks were 1.6 times more likely to give birth to a macrosomic infant, as compared with mothers with a gestational age of 40 weeks. The adjusted odds ratios for maternal age and parity indicated that these variables were weakly associated with macrosomia. The overall mean infant birth weight predicted from the multiple linear regression model was 3441 gm (Table III). A progressive increase in infant birth weight was predicted as maternal BMI increased. After adjustment for maternal height, ethnicity, smoking status, maternal age, parity, gestational age, and infant sex, severely overweight mothers had infants who weighed 156 gm more than infants born to mothers in the reference group for the 25th to <75th percentile. Infants born to the group of tallest mothers were predicted to weigh 222 gm more than those born to the group of shortest mothers. Smoking also had an important effect on birth weight: the infants born to smoking mothers were predicted to weigh 209 gm less than those born to nonsmokers. Infants with the oldest gestational age of 42 weeks were predicted to weigh 78 gm more than infants with a gestational age of 40 weeks. In addition, we found that both maternal age and parity affected infant birth weight.
Volume 162 Number 2
Comment
Pregravid obesity had an important influence on macrosomia among this low-income population. After adjustment for factors known to be important predictors of birth weight, the odds of delivering a macrosomic infant were 2.2 times higher for severely overweight mothers than those for mothers of normal weight. This study identified a 17% prevalence of macrosomia among severely overweight mothers. A previous study of a low-income population found similar levels of macrosomia among children of obese mothers"; however, the different standard used to define obesity in this stud y makes exact comparison difficult. Harrison et al. 11 found an even higher prevalence of macrosomia (approximately 33%) among his group of massively obese mothers (those with > 150% of median weight-forheight according to the National Academy of Sciences weight references, 1974). Our results indicate that the coexistence of additional factors such as greater maternal height, older gestational age, and nonsmoking status had important independent effects on the prevalence of macrosomia in our study. Future research should focus on maternal height and maternal BMI as important variables in the determination of birth weight and infant outcome. In addition to examining the prevalence of macrosomia, we researched the association between maternal overweight and mean infant birth weight. The association we observed between higher mean infant birth weight and maternal obesity is comparable to some, but not all, of the previously published studies. Edwards et al. 6 observed that infants born to massively obese mothers (those with body weights> 150% of Metropolitan Life Insurance standards) weighed an average of 209 gm more than infants born to nonobese control mothers (i.e., those with < 120% of ideal body weight). Previous research with low-income populations '2 . 13 did not identify an association between maternal obesity and mean infant birth weight. In a study of 86 women enrolled in a WIC program, George et al. 13 reported no association between maternal weight-for-height (with the Metropolitan Life Insurance standards) and birth weight. Likewise, Haworth et al. 12 reported no significant increase in birth weight when maternal weight increased among a group of 153 public patients. However, the power of these two studies of low-income populations was limited by their small sample sizes. Maternal height was an important independent risk factor for higher birth weight. Previous studies have shown conflicting results of the influence of maternal height on birth weight. An earlier study" indicated that height had an insignificant role in the determination of birth weight after adjustment for maternal weight. Winikoff and Debrovner 15 showed that among women categorized as high weight-for-height, maternal height
Influence of maternal overweight 493
Table III. Mean birth weight for selected variables PredIcted bIrth welght* (gm)
Gram difference from reference group (95% el)
Body mass index (NHANES II percentile) 3337 - 104 (-110 to - 98) :S25th 25th to <75th 3441 0 75th to <85th 3503 62 (54 to 70) 85th to <95th 3532 91 (83 to 107) 2:95th 3597 156 (147 to 166) Maternal height (quartiles) o First 3441 Second 3516 75 (67 to 82) 144 (137 to 151) Third 3585 Fourth 3663 222 (214 to 230) Ethnic group 3441 White 0 Black 3245 - 196 (- 20 1 to - 190) 3413 - 28 (- 36 to - 19) Hispanic Smoking 3441 No 0 Yes -209 (-214 to -203) 3232 Maternal age (yr) -9 (-15 to -2) 3432 15-20 3441 20-24 0 3449 8 (l to 15) 25-30 2:30 3479 38 (29 to 46) Previous live births 0 o 3441 3495 54 (48 to 61) 1 3512 71 (63 to 79) 2 2:3 3525 84 (75 to 93) Gestational age (wk) - 286 (- 295 to - 276) 3155 37 -175 (-184 to -168) 3266 38 -74 (-81 to -67) 3367 39 3441 0 40 66 (58 to 74) 41 3507 78 (69 to 88) 42 3519 Infant sex Male 0 3441 -135 (-140 to -131) 3306 Female *Multiple linear regression with BMI, height, ethnicity, smoking, maternal age, previous live births. gestational age, and infant sex in the model.
had a significant influence on birth weight, but no such influence was observed for mothers of intermediate or low weight-for-height. Data from the collaborative perinatal stud y '6 support our findings by showing that white mothers who were >67 inches tall had infants that weighed almost 300 gm more than those mothers who were 58 to 60 inches tall. Tall black mothers in the collaborative perinatal study '6 gave birth to infants who weighed 219 gm more than those babies born to mothers 58 to 60 inches tall. Our results suggest that taller mothers may be more prone to having a macrosomic infant; whether these infants have the same risk of adverse outcome as macrosomic infants of shorter mothers needs to be addressed in future studies. Several factors may limit the generalizability of this
494
Larsen, Serdula, and Sullivan
study to other low-income groups. Use of a WIC population for the analysis may limit the application of the data to other low-income groups. The WIC program has been documented to result in a 31 gm increase in birth weight,I7 although other studies do not show a significant effect. 18 Another major limitation of the study is the uncertain accuracy of the data items because of self-reporting of information. Heavier females tend to underreport their weight. 19 Self-reporters also tend to state that they are slightly taller than their actual height. 19 Consequently, self-reporting may underestimate the actual prevalence of overweight and may overestimate height. These inaccuracies in self-reporting might lead researchers to misclassify mothers into a lower BMI category; the result would be an underestimation of the effect of BMI on birth weight. The accuracy of maternally reported birth weights may also be a study limitation. However, Gayle et aJ.2° have shown a high level of concordance between maternally reported birth weights from a WIC clinic and birth weights recorded on birth certificates. Gestational age was assessed at the first prenatal WIC visit by the estimated date of confinement or by the maternal recollection of the last menstrual period, and was not confirmed by ultrasonography or physical examination of the infant. Classification of smoking status was recorded at the first WIC visit and may not have reflected either the quantity or duration of smoking throughout the pregnancy. Other factors that limit this study are a lack of data on weight gain in pregnancy and a lack of information on the prevalence of disease entities that may independently influence birth weight, such as gestational diabetes, hypertension, and fetal anomalies. Particularly gestational diabetes may have an effect on birth weight and may be more common in obese subjects. The high prevalence of obesity among this and other low-income populations in the United States has aroused concern.7 Low-income WIC clients are more likely to be overweight than are private patients," but this association may be partly attributed to the WIC program's use of obesity as an enrollment criterion. 17 Health risks from obesity, such as hypercholesterolemia, diabetes, and hypertension,7 persist after delivery and contribute to long-term morbidity and mortality. Furthermore, infants of obese mothers may also become obese infants by 1 year of age. 6 Additional research should be conducted to determine whether the morbidity associated with macrosomia can be attributed to macrosomia per se or to the associated problems influenced by obesity. In conclusion, pregravid overweight in this population had a significant independent effect on the prevalence of macrosomia. Postpartum weight loss should be emphasized by practitioners and by the WIC pro-
February 1990 Am J Obstet Gynecol
gram to prevent future maternal and infant morbidity. Efforts designed to minimize the prevalence of macrosomia should focus on low-income women because of their high prevalence of obesity and limited access to health care. REFERENCES 1- Spellacy WN, Miller S, Winegar A, Peterson PQ. Macrosomia-maternal characteristics and infant complications. Obstet Gynecol 1985;66:158-612. Stevenson DK, Hopper AO, Cohen RS, Bucalo LR, Kerner JA, Sunshine P. Macrosomia: causes and consequences. J Pediatr 1982;100:515-20. 3. National Center for Health Statistics. Vital health statistics of the United States, 1965, Volume I, Natality. Washington, DC: National Center for Health Statistics, 1967:1-27. 4. National Center for Health Statistics. Vital health statistics of the United States, 1984, Volume I, Natality. Hyattsville, Maryland: National Center for Health Statistics, 1988:68; DHHS publication no. (PHS) 88-1100. 5. Calandra C, Abell DA, Beischer NA. Maternal obesity in pregnancy. Obstet Gynecol 1981;57:8-12. 6. Edwards LE, Dickes WF, Alton IR, Hakanson EY. Pregnancy in the massively obese: course, outcome, and obesity prognosis of the infant. AM J OBSTET GYNECOL 1978; 131:479-83. 7. van Itallie TB. Health implications of overweight and obesity in the United States. Ann Intern Med 1985;103: 983-8. 8. Najjar MF, Rowland M. Anthropometric reference data and prevalence of overweight, United States, 1976-80. Hyattsville, Maryland: National Center for Health Statistics, 1987; DHHS publication no. (PHS) 87-1688, (Vital and health statistics; series 11; no 238). 9. SAS Institute Inc. SAS user's group international supplemental library user's guide, version 5 ed. Cary, North Carolina: SAS Institute Inc., 1986:269-93. 10. SAS Institute Inc. SAS user's guide: statistics, version 5 ed. Cary, North Carolina: SAS Institute Inc., 1985:655709. 11. Harrison GG, Udall IN, Morrow G. Maternal obesity, weight gain in pregnancy, and infant birth weight. AM J OBSTET GYNECOL 1980;136:411-2. 12. Haworth JC, Ellestad-Sayed JJ, King J, Dilling LA. Relation of maternal cigarette smoking, obesity, and energy consumption to infant size. AM J OBSTET GYNECOL 1980; 138: 1185-9. 13. George NN, Kim SK, Duhring JL. Prepregnancy weights and weight gains related to birth weights of infants born to overweight women. J Am Diet Assoc 1984;84:450-2. 14. Eastman NJ, Jackson E. Weight relationships in pregnancy. I. The bearing of maternal weight gain and prepregnancy weight on birth weight in full term pregnancies. Obstet Gynecol Surv 1968;23: 1003-24. 15. Winikoff B, Debrovner CH . Anthropometric determinants of birth weight. Obstet Gynecol 1981 ;58:678-84. 16. Niswander KR, Gordon M. The women and their pregnancies. Washington, DC: National Institutes of Health, 1972:81; DHEW publication no (NIH) 73-379. 17. Schramm WF. Prenatal participation in WIC related to medicaid costs for Missouri newborns: 1982 update. Public Health Rep 1986;101:607-15. 18. Rush D, Sloan NL, LeightonJ, et al. V. Longitudinal study of pregnant women. Am J Clin Nutr 1988;48:439-83. 19. Stewart AL. The reliability and validity of self-reported weight and height. J Chronic Dis 1982;35:295-309. 20. Gayle HD, Yip R, Frank MJ, Nieburg P, Binkin NJ. Validation of maternally reported birth weights among 46,637 Tennessee WIC program participants. Public Health Rep 1988;103:143-7.
tus and uterine activity. At present it seems prudent to educate all pregnant women about these prodromal preterm labor symptoms, much as is done for the equally nonspecific common early symptoms of preeclampsia. REFERENCES
February 1990 Am J Obstet Gynecol
2. Creasy RK. Preterm labor and delivery. In: Creasy RK, Resnik R, eds. Maternal-fetal medicine principles and practice. Philadelphia: WB Saunders, 1984:415. 3. March of Dimes Birth Defects Foundation. Prevention of preterm labor and low birth weight package. White Plains, New York: March of Dimes, 1983. 4. Newman RB, Gill PJ, Wittreich P, Katz M. Maternal perception of prelabor uterine activity. Obstet Gynecol 1986; 68:765-9.
1. Herron MA, Katz M, Creasy RK. Evaluation of a preterm birth prevention program: preliminary report. Obstet Gynecol 1982;59:452-6.
Macrosomia: Influence of maternal overweight among a low-income population Christine E. Larsen, MD, MS, Mary K. Serdula, MD, MPH, and Kevin M. Sullivan, MPH Atlanta, Georgia Macrosomia (birth weight >4000 gm) is associated with a wide variety of adverse intrapartum and perinatal outcomes. To evaluate the effect of pregravid obesity on infant birth weight, we examined data from a low-income population of women (n = 127,879). The population was divided into five groups on the basis of pregravid body mass index (weight/heighf') designated by the National Health and Nutrition Examination Survey II reference population «25th percentile, 25th to <75th percentile, 75th to <85th percentile, 85th to <95th percentile, and 2!95th percentile). The prevalence of infant macrosomia ranged from 5% for the lowest group to 17% for the highest group. With the use of the second group (25th to <75th percentile) as a reference, odds ratios (adjusted for maternal age, smoking status, race, height, parity, gestational age, and infant sex) for macrosomia for the five sequential weight groups were 0.6, 1.0, 1.3, 1.6 and 2.2. We conclude that pregravid overweight had a significant independent effect on birth weight outcome. (AM J OSSTET GVNECOL 1990;162:490-4.)
Key words: Fetal macrosomia, birth weight, obesity
Macrosomia (birth weight >4000 gm) is associated with increased risk of infant birth trauma, fetal death, and cesarean deliveries. l Early recognition of maternal factors that contribute to macrosomia may decrease the incidence of birth injuries. 2 Because the prevalence of macrosomia increased in the U.S. population from 8.2% in 1965' to 11.1 % in 1984: the potential morbidity from macrosomia may escalate. Obese gravidous women have higher levels of hypertension, hyperglycemia, prolonged labor postamniotomy, postpartum hemorrhage, and puerperal pyrexia than do pregnant women who are not obese." From the Division of Nutrition, Center for Health PromotIOn and Education, Centers for DISease Control. Received for publication November 28, 1988; revised May 26, 1989; accepted August 7, 1989. Reprint requests: Mary K. Serdula, MD, MPH, DzvlSlon of Nutrition, Centers for DISease Control, Mazlstop A-41, Atlanta, GA
30333. 611/15881
490
Because the prevalence of macrosomia is higher among infants born to obese mothers than among infants born to nonobese mothers,6 macrosomia is of particular concern in low-income populations in which obesity is a major health problem. 7 Furthermore, because lowincome populations may have limited access to medical care, they may incur even greater risks associated with macrosomia. Although the association between maternal weight and macrosomia has been well documented, the observed association may have been a result of factors other than maternal overweight per se. Previous studies that examined the association between maternal obesity and mauosomia have not adequately accounted for other variables related to macrosomia. In addition, only limited research is available on the prevalence and risk factors for macrosomia among low-income populations. Our study evaluated the influence of pregravid
Influence of maternal overweight 491
Volume 162 Number 2
Table I. Characteristics of study population by ethnic group (%) WhIte
n
= 64,677
Black
n
Body mass index (NHANES II percentiles) 28.0 :S:25th 25th to <75th 43.3 75th to <85th 9.5 11.4 85th to <95th 2:95th 7.8 Maternal height (quartiles) 15.9 First 27.3 Second Third 29.1 27.8 Fourth 40.9 Smoking (yes) Maternal age (yr) 39.1 15-20 29.9 20-25 23.1 25-30 8.0 2:30 Previous live births 43.8 0 30.5 I 16.1 2 2:3 9.7 Gestational age (wk) 7.8 37 14.3 38 22.6 39 26.2 40 18.9 41 10.2 42
weight on birth weight outcome among a low-income population attending the Special Supplemental Food Programs for Women, Infants, and Children (WIC). We analyzed this influence after adjustment for the effects of smoking status, ethnicity, maternal height, maternal age, gestational age, and parity. Material and methods
Through its Pregnancy Nutrition Surveillance System, the Division of Nutrition of the Centers for Disease Control receives information from women who attend the WIC program. WIC records consist of maternal and infant information obtained at prenatal visits and at a postpartum visit. For this study we analyzed data from term (37 through 42 weeks' gestation), singleton live births born between 1983 and 1987. Infants weighing < 1000 gm or >6000 gm were excluded from the data analysis to minimize the likelihood of coding errors. Any records with missing or out-of-range values for variables used in this analysis were excluded. Only records from the white, black, and Hispanic ethnic groups were included. These exclusions left 127,879 records from 14 states. Data items were recorded by WIC clinic personnel. All data entries comprised information provided by the WIC client. No information was available on the prevalence of gestational diabetes, hypertension, and fetal anomalies. We examined the effects of pre gravid body mass in-
= 47,237
n
Hispanzc
= 15,965
n
=
Total
127,879
20.7 44.5 11.8 13.8 9.3
17.4 47.7 14.0 14.6 6.2
24.0 44.3 10.9 12.7 8.2
16.4 26.4 28.1 29.1 20.8
44.5 30.6 17.0 8.0 1l.8
19.7 27.4 27.2 25.8 29.9
36.3 27.8 25.2 10.7
26.0 27.9 28.9 17.3
36.4 28.8 24.6 10.2
33.0 32.1 19.0 15.9
25.2 35.2 21.6 18.1
37.5 31.7 17.9 13.0
Il.l
8.5 16.6 24.8 26.6 15.5 7.9
9.1 16.0 23.7 25.8 16.6 8.8
18.0 24.9 24.9 13.9 7.3
dex (BMI) on both infant macrosomia and mean infant birth weight. The BMI was the independent variable used in this study and was calculated by dividing selfreported prepregnant weight (kilograms) by height 2 (meters squared). The population was grouped into five categories on the basis of the percentile distribution of BMIs from the National Health and Nutrition Examination Survey (NHANES) II reference sample for women ages 20 to 29 years: <25th percentile «20.0 BMI units), 25th to <75th percentile (2:20.0 to 24.9 BMI units), 75th to <85th percentile (2:24.9 to 27.3 BMI units), 85th to <95th percentile (2:27.3 to 32.3 BMI units), and 2:95th percentile (2:32.3 BMI units)." The reference group used for statistical comparison was the 25th to <75th percentile group from NHANES II. This group most closely approximates an ideal body weight category. Women with BMIs in the 85th to <95th percentile were categorized as overweight, and those in the 2:95th percentile were categorized as severelyoverweight. Maternal height was divided into the following approximate quartiles: first (:s: 157.0 cm), second (157.1 to 162.2 cm), third (162.3 to 167.3 cm), and fourth (2: 167.4 cm). Smoking status (yes, no) was reported by the mother on her first prenatal WIC visit. Infant birth weight was reported by the mother at the postpartum visit. Gestational age was calculated on a weekly basis from the self-reported estimated date of confinement and the infants' birthdate. If the estimated
492 larsen, Serdula, and Sullivan
February 1990 Am J Obstet Gynecol
Table II. Prevalence of macrosomia and odds ratios for selected factors Macrosomic infants (%)
Adjusted odds ratio (95% Cl)*
Body mass index (NHANES II percentile) s25th 4.8 0.58 (0.55-0.62) 25th to <75th 8.1 LOOt 75th to <85th 10.8 1.34 ( 1.26-1.43) 85th to <95th 12.8 1.61 (1.52-1.70) ~95th 17.0 2.15 (2 .01-2.29) Maternal height (quartiles) First 5.9 1.00t Second 7.4 1.41 ( 1.32-1.52) Third (1.81-2.07) 9.3 1.93 Fourth 12.5 2.81 (2.63-3.01) Ethnic group White 11.1 LOOt Black 5.6 0.38 (0.36-0.40) Hispanic 10.1 0.86 (0.81-0.92) Smoking No 10.5 LOOt Yes 5.2 0.40 (0.38-0.42) Maternal age (yr) 15-19 6.9 (0.88-0.99) 0.93 20-24 8.7 LOOt 25-29 10.3 1.09 (1.03-1.15) ~30 13.5 1.33 ( 1.25-1.42) Previous live births 0 7.3 LOOt I 8.8 (1.18-1.32) 1.25 10.3 2 1.41 ( 1.32-1.50) ~3 12.0 1.54 (1.43-1.65) Gestational age (wk) 37 3.9 0.38 (0.34-0.42) 38 5.1 0.49 (0.46-0.53) 39 7.1 0.69 (0.65-0.73) 40 10.1 1.00t 41 13.2 1.34 (1.26-1.41 ) 42 14.7 1.55 (1.45-1.66) Infant sex Male 11.2 LOt Female 6.7 0.52 (0.50-0.54) *Multiple logistic regression with BMI, height, ethnicity, smoking, maternal age, previous live births, gestational age, and infant sex in the model. tReference group. date of confinement was not available, then gestational age was calculated from the maternally reported last menstrual period. In the univariate analysis, we computed the prevalence of macrosomia by BMI category. Our multiple logistic regression models used macrosomia as the dependent variable to adjust for the mother's BMI (five categories), height (quartiles), smoking status (yes, no), previous live births (0, 1, 2, ~3), maternal age (15 to 19, 20 to 24, 25 to 29, ~30 years), ethnicity (white, black, Hispanic), gestational age (37, 38, 39, 40, 41 , 42 weeks), and infants' sex. The magnitude and direction of associations between macrosomia and the independent variables were estimated by the odds ratio. The effects of maternal BMI on mean birth weight were quantified with multiple linear regression 'o according Q
to the same procedures as described for the logistic model. Results
Of the sample population, 51 % were white, 37% were black, and 12% were Hispanic (Table I). The prevalence of smoking varied from 41 % for whites to 21 % for blacks and 12% for Hispanics. The prevalences for mothers with BMIs ~85th percentile of the NHANES II reference population were 19% for whites, 23% for blacks, and 21 % for Hispanics. As the maternal BMI increased, the prevalence of macrosomia also increased (Table II). The prevalence of macrosomia ranged from 5% for those <25th percentile to 17% for those in the ~95th percentile. After adjustment for maternal height, ethnicity, smoking status, maternal age, parity, gestational age, and infant sex we also found a statistically significant increase in the odds ratios when the BMIs increased. Specifically, with the NHANES II category for the 25th to <75th percentile as a reference group, we found that the adjusted odds ratios of macrosomia for the five sequential weight groups were 0.6, 1.0, 1.3, 1.6 and 2.2 (Table II). In Table II we also present the adjusted odds ratios for macrosomia according to other variables such as maternal height, ethnicity, maternal age, smoking status, parity, gestational age, and infant sex. Maternal height appeared to have a strong influence : the tallest women were three times more likely to have macrosomic infants than were the shortest women. Mothers with a late gestational age of 42 weeks were 1.6 times more likely to give birth to a macrosomic infant, as compared with mothers with a gestational age of 40 weeks. The adjusted odds ratios for maternal age and parity indicated that these variables were weakly associated with macrosomia. The overall mean infant birth weight predicted from the multiple linear regression model was 3441 gm (Table III). A progressive increase in infant birth weight was predicted as maternal BMI increased. After adjustment for maternal height, ethnicity, smoking status, maternal age, parity, gestational age, and infant sex, severely overweight mothers had infants who weighed 156 gm more than infants born to mothers in the reference group for the 25th to <75th percentile. Infants born to the group of tallest mothers were predicted to weigh 222 gm more than those born to the group of shortest mothers. Smoking also had an important effect on birth weight: the infants born to smoking mothers were predicted to weigh 209 gm less than those born to nonsmokers. Infants with the oldest gestational age of 42 weeks were predicted to weigh 78 gm more than infants with a gestational age of 40 weeks. In addition, we found that both maternal age and parity affected infant birth weight.
Volume 162 Number 2
Comment
Pregravid obesity had an important influence on macrosomia among this low-income population. After adjustment for factors known to be important predictors of birth weight, the odds of delivering a macrosomic infant were 2.2 times higher for severely overweight mothers than those for mothers of normal weight. This study identified a 17% prevalence of macrosomia among severely overweight mothers. A previous study of a low-income population found similar levels of macrosomia among children of obese mothers"; however, the different standard used to define obesity in this stud y makes exact comparison difficult. Harrison et al. 11 found an even higher prevalence of macrosomia (approximately 33%) among his group of massively obese mothers (those with > 150% of median weight-forheight according to the National Academy of Sciences weight references, 1974). Our results indicate that the coexistence of additional factors such as greater maternal height, older gestational age, and nonsmoking status had important independent effects on the prevalence of macrosomia in our study. Future research should focus on maternal height and maternal BMI as important variables in the determination of birth weight and infant outcome. In addition to examining the prevalence of macrosomia, we researched the association between maternal overweight and mean infant birth weight. The association we observed between higher mean infant birth weight and maternal obesity is comparable to some, but not all, of the previously published studies. Edwards et al. 6 observed that infants born to massively obese mothers (those with body weights> 150% of Metropolitan Life Insurance standards) weighed an average of 209 gm more than infants born to nonobese control mothers (i.e., those with < 120% of ideal body weight). Previous research with low-income populations '2 . 13 did not identify an association between maternal obesity and mean infant birth weight. In a study of 86 women enrolled in a WIC program, George et al. 13 reported no association between maternal weight-for-height (with the Metropolitan Life Insurance standards) and birth weight. Likewise, Haworth et al. 12 reported no significant increase in birth weight when maternal weight increased among a group of 153 public patients. However, the power of these two studies of low-income populations was limited by their small sample sizes. Maternal height was an important independent risk factor for higher birth weight. Previous studies have shown conflicting results of the influence of maternal height on birth weight. An earlier study" indicated that height had an insignificant role in the determination of birth weight after adjustment for maternal weight. Winikoff and Debrovner 15 showed that among women categorized as high weight-for-height, maternal height
Influence of maternal overweight 493
Table III. Mean birth weight for selected variables PredIcted bIrth welght* (gm)
Gram difference from reference group (95% el)
Body mass index (NHANES II percentile) 3337 - 104 (-110 to - 98) :S25th 25th to <75th 3441 0 75th to <85th 3503 62 (54 to 70) 85th to <95th 3532 91 (83 to 107) 2:95th 3597 156 (147 to 166) Maternal height (quartiles) o First 3441 Second 3516 75 (67 to 82) 144 (137 to 151) Third 3585 Fourth 3663 222 (214 to 230) Ethnic group 3441 White 0 Black 3245 - 196 (- 20 1 to - 190) 3413 - 28 (- 36 to - 19) Hispanic Smoking 3441 No 0 Yes -209 (-214 to -203) 3232 Maternal age (yr) -9 (-15 to -2) 3432 15-20 3441 20-24 0 3449 8 (l to 15) 25-30 2:30 3479 38 (29 to 46) Previous live births 0 o 3441 3495 54 (48 to 61) 1 3512 71 (63 to 79) 2 2:3 3525 84 (75 to 93) Gestational age (wk) - 286 (- 295 to - 276) 3155 37 -175 (-184 to -168) 3266 38 -74 (-81 to -67) 3367 39 3441 0 40 66 (58 to 74) 41 3507 78 (69 to 88) 42 3519 Infant sex Male 0 3441 -135 (-140 to -131) 3306 Female *Multiple linear regression with BMI, height, ethnicity, smoking, maternal age, previous live births. gestational age, and infant sex in the model.
had a significant influence on birth weight, but no such influence was observed for mothers of intermediate or low weight-for-height. Data from the collaborative perinatal stud y '6 support our findings by showing that white mothers who were >67 inches tall had infants that weighed almost 300 gm more than those mothers who were 58 to 60 inches tall. Tall black mothers in the collaborative perinatal study '6 gave birth to infants who weighed 219 gm more than those babies born to mothers 58 to 60 inches tall. Our results suggest that taller mothers may be more prone to having a macrosomic infant; whether these infants have the same risk of adverse outcome as macrosomic infants of shorter mothers needs to be addressed in future studies. Several factors may limit the generalizability of this
494
Larsen, Serdula, and Sullivan
study to other low-income groups. Use of a WIC population for the analysis may limit the application of the data to other low-income groups. The WIC program has been documented to result in a 31 gm increase in birth weight,I7 although other studies do not show a significant effect. 18 Another major limitation of the study is the uncertain accuracy of the data items because of self-reporting of information. Heavier females tend to underreport their weight. 19 Self-reporters also tend to state that they are slightly taller than their actual height. 19 Consequently, self-reporting may underestimate the actual prevalence of overweight and may overestimate height. These inaccuracies in self-reporting might lead researchers to misclassify mothers into a lower BMI category; the result would be an underestimation of the effect of BMI on birth weight. The accuracy of maternally reported birth weights may also be a study limitation. However, Gayle et aJ.2° have shown a high level of concordance between maternally reported birth weights from a WIC clinic and birth weights recorded on birth certificates. Gestational age was assessed at the first prenatal WIC visit by the estimated date of confinement or by the maternal recollection of the last menstrual period, and was not confirmed by ultrasonography or physical examination of the infant. Classification of smoking status was recorded at the first WIC visit and may not have reflected either the quantity or duration of smoking throughout the pregnancy. Other factors that limit this study are a lack of data on weight gain in pregnancy and a lack of information on the prevalence of disease entities that may independently influence birth weight, such as gestational diabetes, hypertension, and fetal anomalies. Particularly gestational diabetes may have an effect on birth weight and may be more common in obese subjects. The high prevalence of obesity among this and other low-income populations in the United States has aroused concern.7 Low-income WIC clients are more likely to be overweight than are private patients," but this association may be partly attributed to the WIC program's use of obesity as an enrollment criterion. 17 Health risks from obesity, such as hypercholesterolemia, diabetes, and hypertension,7 persist after delivery and contribute to long-term morbidity and mortality. Furthermore, infants of obese mothers may also become obese infants by 1 year of age. 6 Additional research should be conducted to determine whether the morbidity associated with macrosomia can be attributed to macrosomia per se or to the associated problems influenced by obesity. In conclusion, pregravid overweight in this population had a significant independent effect on the prevalence of macrosomia. Postpartum weight loss should be emphasized by practitioners and by the WIC pro-
February 1990 Am J Obstet Gynecol
gram to prevent future maternal and infant morbidity. Efforts designed to minimize the prevalence of macrosomia should focus on low-income women because of their high prevalence of obesity and limited access to health care. REFERENCES 1- Spellacy WN, Miller S, Winegar A, Peterson PQ. Macrosomia-maternal characteristics and infant complications. Obstet Gynecol 1985;66:158-612. Stevenson DK, Hopper AO, Cohen RS, Bucalo LR, Kerner JA, Sunshine P. Macrosomia: causes and consequences. J Pediatr 1982;100:515-20. 3. National Center for Health Statistics. Vital health statistics of the United States, 1965, Volume I, Natality. Washington, DC: National Center for Health Statistics, 1967:1-27. 4. National Center for Health Statistics. Vital health statistics of the United States, 1984, Volume I, Natality. Hyattsville, Maryland: National Center for Health Statistics, 1988:68; DHHS publication no. (PHS) 88-1100. 5. Calandra C, Abell DA, Beischer NA. Maternal obesity in pregnancy. Obstet Gynecol 1981;57:8-12. 6. Edwards LE, Dickes WF, Alton IR, Hakanson EY. Pregnancy in the massively obese: course, outcome, and obesity prognosis of the infant. AM J OBSTET GYNECOL 1978; 131:479-83. 7. van Itallie TB. Health implications of overweight and obesity in the United States. Ann Intern Med 1985;103: 983-8. 8. Najjar MF, Rowland M. Anthropometric reference data and prevalence of overweight, United States, 1976-80. Hyattsville, Maryland: National Center for Health Statistics, 1987; DHHS publication no. (PHS) 87-1688, (Vital and health statistics; series 11; no 238). 9. SAS Institute Inc. SAS user's group international supplemental library user's guide, version 5 ed. Cary, North Carolina: SAS Institute Inc., 1986:269-93. 10. SAS Institute Inc. SAS user's guide: statistics, version 5 ed. Cary, North Carolina: SAS Institute Inc., 1985:655709. 11. Harrison GG, Udall IN, Morrow G. Maternal obesity, weight gain in pregnancy, and infant birth weight. AM J OBSTET GYNECOL 1980;136:411-2. 12. Haworth JC, Ellestad-Sayed JJ, King J, Dilling LA. Relation of maternal cigarette smoking, obesity, and energy consumption to infant size. AM J OBSTET GYNECOL 1980; 138: 1185-9. 13. George NN, Kim SK, Duhring JL. Prepregnancy weights and weight gains related to birth weights of infants born to overweight women. J Am Diet Assoc 1984;84:450-2. 14. Eastman NJ, Jackson E. Weight relationships in pregnancy. I. The bearing of maternal weight gain and prepregnancy weight on birth weight in full term pregnancies. Obstet Gynecol Surv 1968;23: 1003-24. 15. Winikoff B, Debrovner CH . Anthropometric determinants of birth weight. Obstet Gynecol 1981 ;58:678-84. 16. Niswander KR, Gordon M. The women and their pregnancies. Washington, DC: National Institutes of Health, 1972:81; DHEW publication no (NIH) 73-379. 17. Schramm WF. Prenatal participation in WIC related to medicaid costs for Missouri newborns: 1982 update. Public Health Rep 1986;101:607-15. 18. Rush D, Sloan NL, LeightonJ, et al. V. Longitudinal study of pregnant women. Am J Clin Nutr 1988;48:439-83. 19. Stewart AL. The reliability and validity of self-reported weight and height. J Chronic Dis 1982;35:295-309. 20. Gayle HD, Yip R, Frank MJ, Nieburg P, Binkin NJ. Validation of maternally reported birth weights among 46,637 Tennessee WIC program participants. Public Health Rep 1988;103:143-7.