Maternal prepregnant body mass index and weight gain related to low birth weight in South Carolina.

Objectives: The primary purpose of this study was to describe the proportion of low birth weight that could be potentially prevented by programs focusing on maternal prepregnant body mass index (BMI) and/or weight gain during pregnancy.

Methods: In this historic cohort design, study data consisted of birth certificates linked to the Pregnancy Risk Assessment Monitoring System for South Carolina resident women delivering in South Carolina during 1998 and 1999. Statistical analysis was conducted with the use of [chi square], population-attributable risk, and logistic regression. The analysis was performed using SUDAAN to accommodate the analysis weight and extrapolate the sample data to the South Carolina state population.

Results: Eight percent of the very low birth weight (VLBW) rate in South Carolina can be attributed to inadequate weight gain in pregnancy. Approximately 19% of the state's VLBW rate can be attributed to either underweight or overweight BMI at conception. Women with less than adequate weight gain were 1.4 times more likely to deliver a VLBW baby and 1.9 times more likely to deliver a moderately low birth weight baby as compared with women with adequate weight gain.

Conclusions: Appropriate maternal BMI at conception followed by adequate weight gain during pregnancy may have a substantial influence on reducing the number of low birth weight deliveries.

Key Words: body mass index, low birth weight, maternal weight gain

**********

Over the past 20 years, the percentage of low birth weight (LBW; <2,500 g birth weight) babies delivered in South Carolina has not decreased, but instead increased from 8.7% in 1980 to 9.6% in 2000 (Figure). Not only are LBW babies at increased risk for death in the first year of life, but they also have higher rates of severe long-term problems such as cerebral palsy, mental retardation, vision and hearing impairment, and poorer school performance. The initial financial costs associated with care of LBW newborn infants are substantial. In 2000, in South Carolina, the 5,353 LBW babies delivered accounted for $127 million in excess hospital facility charges and more than 69,000 excess hospital days. (1) For more than two decades, attempts to reduce the percentage of LBW babies delivered have been unsuccessful, and South Carolina remains among the states with the highest percentage of LBW babies delivered. (2) Because of the intractable nature of LBW in South Carolina and the serious consequences in both immediate and long-term health status as well as economic impact to the individual and to the state, new data and different approaches to this problem are warranted.

[FIGURE OMITTED]

The South Carolina Pregnancy Risk Assessment Monitoring System (PRAMS) provides that opportunity. PRAMS is a postpartum telephone and mail survey of South Carolina resident women who delivered in South Carolina and is cosponsored by the Centers for Disease Control and Prevention and the South Carolina Department of Health and Environmental Control. The sample of women selected for the survey was determined by a systematic stratified sampling strategy that is weighted on the basis of birth weight. Each respondent was assigned an analysis weight that was then used to extrapolate the sample data to the complete South Carolina population. Survey questions cover a wide array of maternal behaviors and experiences during pregnancy. As such, PRAMS provides access to data not otherwise available on a statewide population.

Among the variables collected in PRAMS are questions specific to maternal weight at conception, height, and weight gain at delivery. Several studies have suggested an association between LBW, maternal weight gain, and prepregnant body mass index (BMI). (3) Women who have a lower than optimal weight gain are 2 times as likely to have a LBW infant. (4-7) Several investigators have suggested that the LBW risk among women with poor weight gain is increased if they also reported a low prepregnant BMI. (8-10) Others found that this association persisted across various ethnic groups. (11,12)

The primary purpose of this study was to describe the association of LBW with maternal prepregnant BMI and weight gain for South Carolina. Maternal weight parameters were selected specifically because they are parameters whose risk is potentially ameliorated before or during pregnancy.

Materials and Methods

The Division of Biostatistics, South Carolina Department of Health and Environmental Control, provided the data for this investigation. The data consist of birth certificates linked to the PRAMS for South Carolina resident women delivering in South Carolina during 1998 and 1999. Data analysis was performed with the use of SUDAAN to accommodate the analysis weight and extrapolate the sample data to the South Carolina state population.

The institutional review board approved exempt status for the study. The data were collected after women had delivered, and no identifiers were used in the analysis.

This investigation selected both traditional confounding variables (ethnicity, prenatal care utilization) as well as data uniquely collected by PRAMS (maternal prepregnant weight, height, weight gain from conception to delivery, intendedness of pregnancy, smoking status, Medicaid status, WIC participation, and whether or not the mother was diabetic or hypertensive during pregnancy). The mother was considered a Medicaid participant if she was enrolled throughout her pregnancy. Prenatal care utilization was determined by the Kessner Index and was modified for this investigation by creating a separate category for no care.

Using prepregnant weight and height, women were classified into one of four prepregnant BMI classifications (maternal weight divided by height squared times 700). These categories were based on the Institute of Medicine's classification groupings and included underweight, <19.8 (U-BMI); normal weight, 19.8 to 26.0 (N-BMI); overweight, 26.1 to 29.0 (OW-BMI); and obese, >29.0 (O-BMI). (13) For each prepregnant BMI category, the expected ideal weight gain range for term singletons at delivery was calculated according to the Institute of Medicine 1990 recommendations, as published by Luke et al. (14)

This investigation extended this work to further classify women as having less than adequate (LTA) or adequate (AWG) weight gain if her weight gain at delivery was below or above the lowest expected weight gain for that particular BMI group expected for her gestational age at delivery. First, for each BMI group, the weight gain at 20 weeks was determined from formula (12.6 lbs for U-BMI; 10.6 lbs for N-BMI; 6.8 lbs for OW-BMI; and 3.8 lbs for O-BMI). Next, the lowest total weight gain expected from the weight gain range for women delivering at term, for each BMI, was obtained (for U-BMI, 28 lbs; N-BMI, 25 lbs; OW-BMI, 15 lbs; and O-BMI, 15 lbs). The weight gain expected at 20 weeks was then subtracted from the total weight gain to arrive at weight gain after 20 weeks. This was then divided by 20 to determine the expected weight gain per week after the 20th week of gestation. Weight gain after the 20th week of gestation is linear to term. Each delivery was then categorized as having LTA or AWG adjusted for gestational age and prepregnant BMI.

Instead of regarding LBW as a homogeneous event, babies delivered at very low birth weight (VLBW; 500 to 1,499 g) were distinguished from babies delivered at moderately low birth weight (MLBW; 1,500 to 2,499 g) to consider potential differences in cause. Outcomes were VLBW or MLBW. The association of each was computed separately for prepregnant BMI status and weight gain categories.

Statistical analysis was conducted by using [chi square], population-attributable risk (PAR), and logistic regression with SAS callable SUDAAN. The PAR is a single marker that takes into account both the prevalence of the risk factor and its severity (relative risk). It has the advantage over other risk measures in that it can be used to estimate the population VLBW or MLBW burden that could be attributed to the risk factor. Stated another way, how much of VLBW or MLBW could be attributed to low or high prepregnant BMI and/or poor weight gain?

Results

During 1998 and 1999, there were 87,293 singleton live births (of at least 500 g birth weight) to residents of South Carolina, delivering in South Carolina. Of these, 1,257 (1.4%) were delivered at VLBW and 5,736 (6.7%) were delivered at MLBW. At conception, 16.2% of women were U-BMI, 52.6% N-BMI, 11.5% OW-BMI, and 19.7% O-BMI. In addition, 27.7% of women were classified as having less than adequate weight gain (for their BMI and gestational age at delivery).

Table 1 presents the characteristics of women according to their prepregnant BMI category. Women classified as UBMI were characterized as being more often white, with intended pregnancy, not participating in WIC, LTA weight gain, and 1.9 times greater likelihood of delivering a MLBW baby. Women classified as O-BMI were characterized more often as white, with unintended pregnancy, participating in WIC, AWG, and 1.8 times greater likelihood of delivering a VLBW baby. Women with LTA weight gain were 1.4 times more likely to deliver a VLBW baby and 1.9 times more likely to deliver a MLBW baby as compared with women with AWG.

Table 2 and Table 3 present the independent and combined risks of prepregnant BMI and weight gain for both VLBW and MLBW deliveries. These data suggest that, except for O-BMI, LTA weight gain increases the risk for VLBW. Furthermore, except for OW-BMI, LTA weight gain increases the risk for MLBW. Most dramatically, compared with N-BMI with adequate weight gain, U-BMI with inadequate weight gain are at 4.1 times greater risk for delivering MLBW infants and 2.0 times greater risk for delivering VLBW infants.

Population-attributable risk was obtained by using the adjusted odds ratios for 6 combinations of BMI (4 categories) and adequacy of weight gain (2 categories) with the reference as N-BMI with adequate weight gain. Adjusted odds ratios were obtained by logistic regression, controlling for potential confounders (ethnicity, intendedness of pregnancy, Medicaid status, WIC status, prenatal care utilization, diabetes, and hypertension). The prevalence, adjusted odds ratios, and PAR are presented in Table 4 and Table 5.

These tables indicate that 8.1% of the VLBW rate in South Carolina can be attributed to inadequate weight gain in pregnancy. Approximately 19% of the state's VLBW rate can be attributed to either underweight or overweight BMI at conception. Among the MLBW, 19% could be attributed to inadequate weight gain, and in addition, 16.8% could be attributed to underweight BMI at conception.

Discussion

From this investigation, a significant percentage of women delivering babies in South Carolina had inadequate weight gain during pregnancy or were either underweight or overweight at conception. Furthermore, each of these conditions was independently associated with an increased risk for delivering a low birth weight baby, but, surprisingly, the associations differed by VLBW or MLBW. Women with LTA weight gain were at 1.4 and 1.9 times greater risk for delivering a VLBW or MLBW baby, respectively. Obese women made up the group at greatest risk for delivering a VLBW baby. Women who were underweight at conception were the group at greatest risk for delivering a MLBW baby.

Examination of the PARs indicated that through a combination of prevalence and adjusted risk, approximately 8% of the South Carolina VLBW and 19% of the MLBW rates were attributed to inadequate weight gain in pregnancy. In addition, approximately 8% of the VLBW and 17% of the MLBW rates were attributed to women who were underweight at conception. The use of adjusted risks provides an opportunity to examine the independent effect of weight gain and prepregnant BMI without potential bias for confounders. Combining VLBW and MLBW, 26.8% of low birth weight is attributable to inadequate weight gain. Alternatively, 25.2% of low birth weight is attributable to women underweight at conception. These populations are not completely independent, and, as such, the PAR estimates cannot be added for an estimated cumulative effect. They are, however, indicative of potential interventions that may significantly improve South Carolina's persistent low birth weight rate.

From these data, approximately 50 VLBW babies and 545 MLBW babies may be attributed to inadequate weight gain during pregnancy each year (regardless of prepregnant BMI). Alternatively, approximately 50 VLBW babies and 487 MLBW babies may be attributed to underweight BMI at conception (regardless of weight gain). Eliminating inadequate weight gain could potentially prevent 595 low birth weight births each year and alternatively, ensuring adequate prepregnant maternal weight could potentially prevent 537 low birth weight births each year. Even if such programs were marginally successful, approximately 50%, for example, the low birth weight savings would be approximately 270 to 300 fewer low birth weight births each year. That would reduce South Carolina's low birth weight rate from 8% (based on this investigation) to 7.3% or 7.4% and, at the same time, would result in an estimated $6.8 to $7.6 million savings in facility charges per year.

Despite attempts to the contrary, the low birth weight rate in South Carolina has been intractable for the past decade. Future attempts to reduce the number of high-risk births will have to focus on new methods of prevention if any sustainable improvements are to be realized. The data from this investigation provide insight into two approaches that can be achieved with little expense. Statewide efforts to promote appropriate BMI can have a substantial impact on the overall health of women and particularly so on those who become pregnant. Emphasizing appropriate weight gain during pregnancy may be best underscored through promotion of early and regular prenatal care. Regardless, if future efforts to prevent low birth weight mimic those of the past, then we should expect future successes to also mimic those of the past.

Table 1. Characteristics of singleton live deliveries, by prepregnant
BMI: South Carolina, 1998 to 1999 (a)

               Characteristic     Total     UW-BMI     N-BMI    OW-BMI

Total                             100.0     16.4       53.6     10.9
Ethnicity      White               56.3     18.6       55.7     10.9
               Black               43.7     11.9       49.3     14.3
Intended       Yes                 87.6     17.1       54.3      9.7
               No                  12.4      9.0       50.1     18.4
Medicaid       Yes                 48.6     16.1       50.1     11.7
               No                  51.4     16.5       57.4      9.6
WIC            Yes                 53.2     14.5       49.9     11.9
               No                  46.8     18.6       57.6      9.8
Prenatal care  Adequate            75.2     17.3       52.8     10.0
               Intermediate        18.6     13.1       57.4     12.9
               Inadequate           5.5     14.1       55.1     13.5
               No care              0.8     27.2       37.2      3.1
Weight gain    Adequate            72.3     14.9       52.1     13.6
               LTA                 27.7     19.6       53.9      5.9
Diabetes       Yes                  4.1      3.7       36.4     21.1
               No                  95.9     16.9       54.3     10.5
Hypertension   Yes                  4.8      7.8       36.4     15.2
               No                  95.2     16.8       54.4     10.7
Outcome        VLBW                 1.3     17.7       42.1     13.0
               MLBW                 6.2     26.4       46.3     10.5

               Characteristic   O-BMI    P value

Total                           19.2
Ethnicity      White            19.2     <0.0001
               Black            24.6
Intended       Yes              18.9      0.01
               No               22.5
Medicaid       Yes              22.1      0.07
               No               16.5
WIC            Yes              23.8     <0.0001
               No               14.0
Prenatal care  Adequate         19.2      0.54
               Intermediate     16.6
               Inadequate       17.2
               No care          32.6
Weight gain    Adequate         19.4     <0.0001
               LTA              20.7
Diabetes       Yes              38.8      0.09
               No               17.6
Hypertension   Yes              40.6      0.56
               No               18.1
Outcome        VLBW             27.2     <0.0001
               MLBW             16.7     <0.0001

(a) BMI, body mass index; UW, underweight; N, normal; OW, overweight; O,
obese; LTA, less than adequate; VLBW, very low birth weight; MLBW,
moderately low birth weight.

Table 2. Association of prepregnant BMI and weight gain with very low
birth weight: South Carolina (a)

                            Underweight BMI  Normal BMI  Overweight BMI
                                (16.2%)        (52.6%)      (11.5%)

Less than adequate (27.7%)        1.9            1.6          2.4
Adequate (72.3%)                  1.2            0.8          1.4
VLBW                              1.4            1.0          1.5

                            Obese BMI
                             (19.7%)   VLBW

Less than adequate (27.7%)     1.4      1.7
Adequate (72.3%)               2.0      1.2
VLBW                           1.8      1.4

(a) BMI, body mass index; VLBW, very low birth weight.

Table 3. Association of prepregnant BMI and weight gain with moderately
low birth weight: South Carolina (a)

                            Underweight BMI  Normal BMI  Overweight BMI
                                (16.2%)        (52.6%)      (11.5%)

Less than adequate (27.7%)       19.0            9.3          3.6
Adequate (72.3%)                  7.0            4.6          6.7
MLBW                             11.0            5.9          6.3

                            Obese BMI
                             (19.7%)   MLBW

Less than adequate (27.7%)     6.7      10.3
Adequate (72.3%)               5.3       5.4
MLBW                           5.7       6.7

(a) BMI, body mass index; MLBW, moderately low birth weight.

Table 4. Prevalence, adjusted odds ratios, and population-attributable
risk for VLBW by prepregnant BMI and weight gain: South Carolina (a)

                            Adjusted OR
              Prevalence     (95% CI)         PAR

LTA            27.7%        1.32 (0.99/1.72)   8.1
UW-BMI         16.2%        1.55 (1.15/1.88)   8.3
OW-BMI         11.5%        1.33 (0.99/1.86)   3.5
O-BMI          19.7%        1.39 (1.06/1.83)   7.0
UW-BMI + LTA    5.4%        2.06 (1.26/2.87)   5.4
N-BMI + LTA    14.9%        1.82 (1.22/2.29)  10.9
OW-BMI + LTA    1.6%        2.05 (0.90/4.44)   1.6
O-BMI + LTA     5.7%        1.25 (0.61/1.61)   1.4
O-BMI + AWG    14.0%        1.74 (1.23/2.42)   9.4

(a) OR, odds ratios; PAR, population-attributable risk; VLBW, very low
birth weight; BMI, body mass index: LTA, less than adequate; UW,
underweight; OW, overweight; O, obese; N, normal; AWG, adequate weight
gain.

Table 5. Prevalence, adjusted odds ratios, and population-attributable
risk for MLBW by prepregnant BMI and weight gain: South Carolina (a)

                            Adjusted OR
              Prevalence     (95% CI)         PAR

LTA              27.7%      1.83 (1.41/2.40)  18.7%
UW-BMI           16.2%      2.25 (1.68/3.09)  17.0
OW-BMI           11.5%      0.95 (0.66/1.49)  (b)
O-BMI            19.7%      0.76 (0.57/1.14)  (b)
UW-BMI + LTA      5.4%      4.83 (2.98/7.83)  17.1
N-BMI + LTA      14.9%      1.77 (1.23/2.60)  10.3
OW-BMI + LTA      1.6%      0.28 (0.11/1.83)  (b)
O-BMI + LTA       5.7%      1.09 (0.67/2.13)   0.5

(a) OR, odds ratios; PAR, population-attributable risk; MLBW, moderately
low birth weight; BMI, body mass index; CI, confidence intervals; LTA,
less than adequate; UW, underweight; OW, overweight; O, obese, N,
normal.
(b) Not computed because adjusted OR are less than 1.0.

Acknowledgments

The authors extend special appreciation to the following, who assisted in this project: Myla Ebeling, Division of Pediatric Epidemiology, Medical University of South Carolina; Kristen Helms, MSPH, Regional Program Manager; Mary Kate Powell, MPH, South Carolina PRAMS Project Coordinator; Meg Weis, MPH, South Carolina PRAMS Data Manager; Guang Zhao, PhD, South Carolina PRAMS Project Director; Sylvia J. Sievers, PhD, South Carolina PRAMS Project Coordinator; Division of Biostatistics and Health GIS, PHSIS, South Carolina Department of Health and Environmental Control.

Accepted June 22, 2004.

References

1. Hulsey TC, Hulsey T, Collins Sharp BA, Wagner CL. Newborn hospital survival and financial impact by birth weight and gestational age. J S C Med Assoc 2003;99:9e.

2. South Carolina Perinatal Regionalization Surveillance Data. Report of Perinatal Indicators for 1985-2000. South Carolina Department of Health and Environmental Control. March 2002.

3. Carmichael SJ, Abrams B, Selvin S. The association of pattern of maternal weight gain and length of gestation and risk of spontaneous preterm delivery. Paediatr Perinat Epidemiol 1997;11:392-406.

4. Siega-Riz AM, Adair LS, Hobel CJ. Maternal underweight status and inadequate rate of weight gain during the third trimester of pregnancy increases the risk of preterm delivery. J Nutr 1996;126:146-153.

5. Kramer MS, Coates AL, Michoud MC, et al. Maternal anthropometry and idiopathic preterm labor. Obstet Gynecol 1995;86:744-748.

6. Hickey CA, Cliver SP, McNeal SF, Goldenberg RL. Prenatal weight gain patterns and birth weight among nonobese black and white women. Obstet Gynecol 1996;88:490-496.

7. Centers for Disease Control. Pediatric Nutrition Surveillance System annual Report, Atlanta: US Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Division of Nutrition and Physical Activity, Maternal and Child Health Branch, 1996.

8. Kramer M. Determinants of low birth weight: methodological assessment and meta-analysis. Bull World Health Organ 1987;65:663-737.

9. Arbuckle T, Sherman G, Kawamoto Y, Meyers A. Predictors of birth weight from the nutrition Canada follow-up cohort. Pediatr Perinat Epidemiol 1989;3:115-129.

10. Merchant SS, Momin IA, Sewani AA, Zuberi NF. Effect of prepregnancy body mass index and gestational weight gain on birth weight. J Pakistan Med Assoc 1999;49:23-25.

11. Schieve LA, Cogswell ME, Scanlon KS. An empiric evaluation of the Institute of Medicine's pregnancy weight gain guidelines by race. Obstet Gynecol 1998;91:878-884.

12. Ogunyemi D, Hullett S, Leeper J, Risk A. Prepregnancy body mass index, weight gain during pregnancy, and perinatal outcome in a rural black population. J Matern Fetal Med 1998;7:190-193.

13. Institute of Medicine: Nutrition during pregnancy: weight gain and nutrient supplementation. Washington, DC, National Academy Press, 1990.

14. Luke B, Hediger ML, Nugent C, et al. Body mass index-specific weight gains associated with optimal birth weights in twin pregnancies J Reprod Med 2003;48:217-224.

RELATED ARTICLE: Key Points

* Correcting either underweight or obese prepregnant body mass index could prevent approximately 19% of very low birth weight deliveries in South Carolina.

* Ensuring adequate weight gain during pregnancy could prevent approximately 8% of very low birth weight deliveries in South Carolina.

* Appropriate weight at conception, followed by adequate weight gain during pregnancy, could potentially prevent 500 low birth weight deliveries each year in South Carolina and save $15 million in newborn charges alone.

* The techniques used in this investigation can be readily applied to any of the 31 states participating in the Centers for Disease Control Pregnancy Risk Assessment Monitoring System project.

Thomas C. Hulsey, MSPH, SCD, Diane Neal, PHD, Shana Catoe Bondo, MD, Tara Hulsey, RN, PHD, and Roger Newman, MD

From the Division of Pediatric Epidemiology, the Department of Obstetrics and Gynecology, and the College of Nursing at the Medical University of South Carolina, Charleston, SC.

This study was carried out with Medical University of South Carolina Institutional Review Board approval No. 673 (January 29, 2004).

Reprint requests to Prof. Thomas C. Hulsey, Division of Pediatric Epidemiology, Medical University of South Carolina, 35 Rutledge Avenue, PO Box 250566, Charleston, SC 29425. E-mail: [e-mail]