Pre-pregnancy body mass index, rate of gestational weight gain, and preterm birth among US Pacific Islanders

Abstract

Objective:

To examine the association between rate of gestational weight gain (GWG) and preterm birth (PTB) classified by pre-pregnancy body mass index (BMI) among Pacific Islanders in the United States.

Methods:

Pacific Islander mothers (n=55,975) and singleton infants (22–41 gestational weeks) without congenital anomalies were included, using data from the National Center for Health Statistics (2014–2018). PTB was compared by pre-pregnancy BMI among women in each stratum of rate of GWG using Cox proportional hazards models.

Results:

Compared to mothers with a rate of GWG within the guidelines, mothers with rate of GWG below the guidelines and either pre-pregnancy underweight (adjusted hazard ratios [aHR]=1.84, 95% confidence interval [CI] 1.10–3.06), healthy weight (aHR=1.38, 95% CI 1.15–1.65), obesity I (aHR=1.22, 95% CI 0.97–1.52), or obesity II (aHR=1.43, 95% CI 1.05–1.96) had an increased risk of PTB; mothers with rate of GWG above the guidelines and either pre-pregnancy underweight (aHR=1.57, 95% CI 0.92–2.69) or obesity II (aHR=1.31, 95% CI 0.98–1.76) had an increased risk of PTB.

Conclusions:

The association between rate of GWG below or above the guidelines and PTB differs by pre-pregnancy BMI among Pacific Islanders.

Introduction

Preterm birth (PTB, live birth <37 weeks gestation) is the leading cause of death in children under five(1). Among US singletons born without congenital anomalies, PTB prevalence increased from 7.6% to 8.1%(2) between 2014 and 2018, with Black women (11.2%) having the highest and White women (7.2%) the lowest prevalence(2). Pacific Islander women in the US [PTB prevalence: 11.2%(3)] also have a higher risk of PTB compared to White women(3), but our understanding of risk factors underlying this disparity is limited since Pacific Islanders have been traditionally aggregated with Asian Americans whose demographic characteristics and biomedical risk profiles are quite dissimilar(4).

In addition to social-environmental risks experienced by all minority populations(5), Pacific Islanders have a higher prevalence of obesity compared to others(6), potentially putting them at additional risk of PTB(7). Pacific Islander adults were approximately three times as likely to have obesity than Asian Americans (2014) and were 80% more likely compared to White Americans (2016)(6). In a single prior study, however, no association between obesity and PTB was observed among Pacific Islander women(3). Beyond pre-pregnancy body mass index (BMI), gestational weight gain (GWG) is another important indicator for PTB and consistently modifies pre-pregnancy BMI’s association with PTB (and vice versa)(8, 9). Pre-pregnancy BMI below or above the healthy range(7, 10) and inadequate GWG(11) have been associated with increased risk of PTB. To our knowledge, no study has examined both risk factors in Pacific Islander pregnant women and in general, few studies have estimated associations among PTB, pre-pregnancy body mass index (BMI) and gestational weight gain (GWG) in women with Class II obesity and higher (BMI >=35kg/m²), where more Pacific Islanders are likely to fall compared to other populations. In a recent multi-ethnic meta-analysis(12), for example, all women with BMI over 30 kg/m² were included in one ‘obesity’ category.

Methodologically, inconsistent findings have been shown when using total versus rate of GWG (lb/week): high total GWG is associated with lower risk of PTB, yet the highest rate of GWG has been associated with greater risk(13). Additionally, recent studies have questioned the use of total GWG in analyses of PTB since we cannot confirm whether inadequate total GWG causes PTB, or whether women experiencing PTB do not have enough time for adequate total GWG(14). Therefore, we aimed to estimate the association between pre-pregnancy BMI and PTB, and how the association between the rate of GWG in the second and third trimesters is modified by pre-pregnancy BMI among Pacific Islanders in the US.

Methods

In this population-based retrospective cohort study, we used 2014–2018 birth cohort data files from the US National Center for Health Statistics (NCHS)(15). Data from 2019 onward were not included to avoid any influence of the COVID-19 pandemic. Since no intervention or interaction with participants occurred, and no private information was identifiable in this public dataset, no research ethics board approval was required. Eligible mother-infant dyads were identified based on mothers whose race was listed as Native Hawaiian or Pacific Islander (NHPI, n=61,579, herein referred to as Pacific Islander). We excluded infants with gestational age (GA) <22 or >41 weeks (n=309), plural births (n=1,529), infants with congenital anomalies (n=2,683), and 1,083 blank records where only maternal race was included. In total, n=55,975 Pacific Islander mother-infant dyads were included in our analyses (Figure 1).

Figure 1.

Flow diagram of the included Pacific Islander mother-infant dyads in the US from the 2014–2018 birth cohort data files

Data on infant and maternal characteristics were obtained from birth certificates. Having a PTB was our outcome of interest. For descriptive purposes, PTBs were further classified as extreme (22–27 weeks), very (28–31 weeks), and moderate to late (32–36 weeks) based on WHO definitions(1). GA was determined by obstetric estimate. Pre-pregnancy BMI(16) (underweight, <18.5 kg/m²; healthy weight, 18.5–24.9 kg/m²; overweight, 25.0–29.9 kg/m²; obesity I, 30.0–34.9 kg/m²; obesity II, 35.0–39.9 kg/m²; obesity III, >=40.0 kg/m²) and rate of GWG for the second and third trimesters(17) (lb/week, below, within [underweight: 1.0–1.3 lb/week; healthy weight: 0.8–1.0 lb/week; overweight: 0.5–0.7 lb/week; obesity class I-III: 0.4–0.6 lb/week], above the 2009 Institute of Medicine [IOM] guidelines) were the risk factors of interest in analyses. Due to a lack of trimester-specific information about weight gain, we assumed that in the study population, weight gain was consistent during pregnancy, and therefore the rate of GWG (GWG/week) for the 2^nd and 3^rd trimesters was calculated as $RateofGWG=\frac{TotalGWG-2.75lb}{GA-12weeks}$, where 2.75 lb is the midpoint of the suggested GWG range (1.1–4.4 lb) during the 1^st trimester(17), and 12 completed weeks is the length of the 1^st trimester. To answer the primary research question, we included the interaction between pre-pregnancy BMI and rate of GWG in our analyses, which allowed us to present the data classified by pre-pregnancy BMI.

Based on the available data collected by the US NCHS, we selected variables to describe the analytical sample with respect to birth outcomes and maternal characteristics, including prenatal care utilization, enrollment in a government program and health insurance. Birth weight was classified with two methods: (I) low birth weight (LWB, <2,500 g), normal weight (2,500–4,000 g), and macrosomia (>4,000 g)(18); (II) small-, appropriate-, and large-for-GA, based on the 2017 US birth weight percentiles for singletons(19). Other birth outcomes included five-minute Appearance, Pulse, Grimace, Activity and Respiration (APGAR) score(20) and final route/method of birth. Maternal characteristics included ethnicity (Hawaiian, Guamanian, Samoan, Other Pacific Islander), age (<20 years, 20–34 years, >=35 years), nativity (born in/outside the US), marital status (married, unmarried), highest educational attainment (less than high school completed, high school graduated, some college credit, associate degree or above), death of prior live-born children (yes/no), parity>1, Adequacy of Prenatal Care Utilization Index(21) (inadequate, intermediate, adequate, adequate-plus), enrollment in the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC; yes/no), smoking history (pre-pregnancy, or during 1^st/2^nd/3^rd trimester), medical risk factors (pre-pregnancy diabetes, gestational diabetes mellitus [GDM], pre-pregnancy hypertension, gestational hypertension, hypertension eclampsia, previous PTB), and payment type (Medicaid, private insurance, self-pay, other).

Since there were differences in the socio-demographic characteristics of dyads with missing values for BMI (5.3%) and GWG (6.0%) compared to those with data (Table S1; indicators were generally poorer among those missing data), these two risk factors and parity>1 (<0.1%), WIC enrollment (3.6%), and Adequacy of Prenatal Care Utilization Index (4.7%) were imputed with the maximum likelihood via expectation maximization (EM)(22) method in order to retain more dyads in our study population. Then, categorical variables were rounded to the nearest integer.

Demographic and prenatal characteristics and birth outcomes were summarized and compared among pre-pregnancy BMI groups, since between the two risk factors – BMI groups and rate of GWG categories – the former took precedence in time and was also considered as an effect modifier in modeling the rate of GWG with PTB. Categorical characteristics are expressed as N (%); ${\chi }^2$ or Fisher’s exact tests were used to assess group differences. Continuous characteristics are expressed as median (Q1-Q3, interquartile range); Wilcoxon two-sample tests were used for group difference assessment. We then used Cox proportional hazard regression models to examine associations between characteristics and PTB, with the time variable of GA at birth. Covariates in the univariable models were selected based on previous literature (3, 8, 23), and covariates retained in the multivariable model were selected based on their clinical relevance and based on the magnitude of their effect size (hazard ratio) and surrounding 95% Confidence Interval (95% CI). Marital status was not retained in models since this information was not available for California birth records (one of the states with the largest population of Pacific Islander residents), which may have biased study findings(15). Proportional hazards assumption was evaluated using the standardized score process, which is based on the cumulative sums of martingale-based residuals (24).Since the interaction between pre-pregnancy BMI and rate of GWG was specified a priori, we retained this interaction in the multivariable model. Interactions between pre-pregnancy BMI and GDM, and between pre-pregnancy BMI and gestational hypertension were also explored in the models, since pre-pregnancy BMI may modify associations between these medical risk factors and PTB(25). To further avoid the potential overadjustment, we also conducted a sensitivity analysis with gestational diabetes, gestational hypertension, and hypertension eclampsia excluded from the adjusted model.

Unadjusted and adjusted hazard rate ratios (aHR) with 95% CI are presented. We also plotted the predicted probability of PTB by GA at birth classified in each pre-pregnancy BMI stratum, including two plots for both the adjusted (multivariable) model and the nested model (only including pre-pregnancy BMI, rate of GWG, and their interaction) outcomes. Our interpretation and conclusions are based on both the magnitude of effect sizes and the coverage of 95% CIs. For example, a wide 95%CI which covers approximately equally likely effects in either direction would be aligned with absence of an association. A 95%CI mostly covering effects in one direction and suggesting on average at least a 10% reduction, or conversely a 10% increase, in the rate of PTB (hazard rate from the Cox model) aligns itself with a meaningful association. Analyses were performed using RStudio (RStudio, Inc.) and SAS version 9.4 (SAS Institute Inc.).

Results

Among the 55,975 Pacific Islander live born singletons without congenital anomalies the average PTB prevalence was 9.3%. Among the included Pacific Islander mothers, 1.8% had a pre-pregnancy BMI indicative of underweight (the fewest), 26.4% had a healthy weight, 29.4% had overweight (the most), 22.3% were classified as having obesity I, 11.8% were classified as having class II obesity, and the remaining 8.2% had class III obesity; 25.7% and 62.3% had rates of GWG below or above the guidelines, respectively (Table 1 [by pre-pregnancy BMI], Table S2 [by rate of GWG]). With the exception of infant sex variable, all other characteristics were associated with pre-pregnancy BMI. The proportions of LBW, SGA, mothers born in the US, married mothers, and spontaneous birth decreased as pre-pregnancy BMI increased. With each increment of pre-pregnancy BMI, the proportion of dyads with the following characteristics increased: infants with macrosomia or LGA; mothers identifying as Samoan, having advanced maternal age, death of a prior child, parity>1, WIC enrolled, history of smoking, pre-pregnancy diabetes, GDM, pre-pregnancy hypertension, gestational hypertension, hypertension eclampsia, previous PTB, and Medicaid insured. There was a u-shaped association between pre-pregnancy BMI and rate of GWG: the proportion of mothers with a rate of GWG within the guidelines decreased from underweight to obesity I, and then increased from obesity I to obesity III. The highest proportion of 9–10 APGAR scores were observed among mothers with a pre-pregnancy BMI of healthy weight.

Table 1.

Characteristics and Birth Outcomes by Pre-pregnancy BMI of Pacific Islander Mother-Infant Dyads in the US, 2014–2018

Characteristics	Underweight (n=1,030, 1.8%)	Healthy weight (n=14,787, 26.4%)	Overweight (n=16,455, 29.4%)	Obesity I (n=12,491, 22.3%)	Obesity II (n=6,609, 11.8%)	Obesity III (n=4,603, 8.2%)	P-value*
	N or median (% or Q1-Q3)	N or median (% or Q1-Q3)	N or median (% or Q1-Q3)	N or median (% or Q1-Q3)	N or median (% or Q1-Q3)	N or median (% or Q1-Q3)
Infant
Sex
Female	525 (51.0)	7,292 (49.3)	8,060 (49.0)	6,010 (48.1)	3,173 (48.0)	2,260 (49.1)
Male	505 (49.0)	7,495 (50.7)	8,395 (51.0)	6,481 (51.9)	3,436 (52.0)	2,343 (50.9)	0.163
Gestational age (weeks)†	38.2 (36.8–39.7)	38.5 (37.3–39.8)	38.5 (37.2–39.9)	38.5 (37.2–39.9)	38.5 (37.1–39.9)	38.5 (37.1–39.8)	<.001
Preterm birth categories
Extreme PTB (22–27 weeks)	7 (0.7)	52 (0.4)	71 (0.4)	69 (0.6)	41 (0.6)	27 (0.6)
Very PTB (28–31 weeks)	13 (1.3)	104 (0.7)	133 (0.8)	87 (0.7)	52 (0.8)	35 (0.8)
Moderate to Late PTB (32–36 weeks)	110 (10.7)	1,110 (7.5)	1,332 (8.1)	1,028 (8.2)	522 (7.9)	399 (8.7)
Term birth (37–41 weeks)	900 (87.4)	13,521 (91.4)	14,919 (90.7)	11,307 (90.5)	5,994 (90.7)	4,142 (90.0)	0.001
Birth weight categories (I)
Low birth weight (<2500 g)	122 (11.8)	1,086 (7.3)	1,077 (6.5)	720 (5.8)	391 (5.9)	241 (5.2)
Normal weight (2500–4000 g)	882 (85.6)	13,061 (88.3)	14,136 (86.0)	10,188 (81.6)	5,181 (78.4)	3,484 (75.7)
Macrosomia (>4000 g)	26 (2.5)	640 (4.3)	1,242 (7.5)	1,583 (12.7)	1,037 (15.7)	878 (19.1)	<.001
Birth weight categories (II)
Smaller for gestational age	201 (19.5)	2,064 (14.0)	1,719 (10.4)	915 (7.3)	416 (6.3)	233 (5.1)
Appropriate for gestational age	786 (76.3)	11,792 (79.8)	13,105 (79.6)	9,476 (75.9)	4,833 (73.1)	3,234 (70.3)
Larger for gestational age	43 (4.2)	931 (6.3)	1,631 (10.0)	2,100 (16.8)	1,360 (20.6)	1,136 (24.7)	<.001
Maternal
Ethnicity
Hawaiian	114 (11.1)	1,694 (11.5)	1,487 (9.0)	1,026 (8.2)	499 (7.6)	328 (7.1)
Guamanian	156 (15.2)	2,521 (17.1)	2,046 (12.4)	1,274 (10.2)	582 (8.8)	319 (6.9)
Samoan	58 (5.6)	1,347 (9.1)	2,636 (16.0)	2,946 (23.6)	2,106 (31.9)	1,924 (41.8)
Other Pacific Islander	702 (68.2)	9,225 (62.4)	10,286 (62.5)	7,245 (58.0)	3,422 (51.8)	2,032 (44.2)	<.001
Maternal age (y)†	25.2 (21.5–28.9)	26.5 (22.6–30.4)	27.6 (23.7–31.5)	28.7 (24.8–32.6)	29.0 (25.2–32.8)	29.6 (25.9–33.2)	<.001
< 20 years	120 (11.7)	1,528 (10.3)	1,095 (6.7)	506 (4.1)	191 (2.9)	94 (2.0)
20–34 years	838 (81.4)	11,776 (79.6)	13,145 (79.9)	9,802 (78.5)	5,260 (79.6)	3,607 (78.4)
>= 35 years	72 (7.0)	1,483 (10.0)	2,215 (13.5)	2,183 (17.5)	1,158 (17.5)	902 (19.6)	<.001
Mother’s nativity
Born in the US	333 (32.3)	5,102 (34.5)	5,581 (33.9)	4,793 (38.4)	2,901 (43.9)	2,337 (50.8)
Born outside the US	697 (67.7)	9,685 (65.5)	10,874 (66.1)	7,698 (61.6)	3,708 (56.1)	2,266 (49.2)	<.001
Marital status‡
Married	398 (40.9)	6,343 (45.5)	7,384 (47.8)	6,106 (52.8)	3,443 (56.8)	2,462 (59.4)
Unmarried	575 (59.1)	7,588 (54.5)	8,087 (52.2)	5,450 (47.2)	2,623 (43.2)	1,686 (40.7)	<.001
Mother’s education
Under high school graduation	308 (29.9)	3,903 (26.4)	4,588 (27.9)	3,058 (24.5)	1,493 (22.6)	941 (20.4)
High school graduated	370 (35.9)	5,006 (33.9)	5,714 (34.7)	4,631 (37.1)	2,569 (38.9)	1,886 (41.0)
Some college credit (no degree)	212 (20.6)	3,281 (22.2)	3,679 (22.4)	2,967 (23.8)	1,574 (23.8)	1,103 (24.0)
Associate degree or above	140 (13.6)	2,597 (17.6)	2,474 (15.0)	1,835 (14.7)	973 (14.7)	673 (14.6)	<.001
Death of prior live-born children	8 (0.8)	204 (1.4)	240 (1.5)	234 (1.9)	131 (2.0)	91 (2.0)	<.001
Parity>1	550 (53.4)	9,003 (60.9)	11,397 (69.3)	9,502 (76.1)	5,106 (77.3)	3,606 (78.3)	<.001
Adequacy of Prenatal Care Utilization Index
Inadequate	340 (33.0)	4,808 (32.5)	6,067 (36.9)	4,510 (36.1)	2,205 (33.4)	1,583 (34.4)
Intermediate	122 (11.8)	1,697 (11.5)	1,718 (10.4)	1,335 (10.7)	744 (11.3)	535 (11.6)
Adequate	358 (34.8)	5,066 (34.3)	5,494 (33.4)	4,027 (32.2)	2,109 (31.9)	1,353 (29.4)
Adequate Plus	210 (20.4)	3,216 (21.8)	3,176 (19.3)	2,619 (21.0)	1,551 (23.5)	1,132 (24.6)	<.001
Enrolled in WIC program	503 (48.8)	7,141 (48.3)	8,142 (49.5)	6,401 (51.2)	3,493 (52.9)	2,537 (55.1)	<.001
Ever smoked	58 (5.63)	756 (5.11)	896 (5.4)	784 (6.28)	501 (7.58)	400 (8.69)	<.001
Rate of GWG (lb/week)
Below the IOM guidelines	437 (42.4)	4,681 (31.7)	3,275 (19.9)	2,640 (21.1)	1,840 (27.8)	1,494 (32.5)
Within the IOM guidelines	240 (23.3)	2,071 (14.0)	2,046 (12.4)	1,191 (9.5)	694 (10.5)	508 (11.0)
Above the IOM guidelines	353 (34.3)	8,035 (54.3)	11,134 (67.7)	8,660 (69.3)	4,075 (61.7)	2,601 (56.5)	<.001
Total GWG (lb)	31.4 (23.0–41.0)	30.0 (21.0–40.0)	27.3 (19.0–39.0)	26.0 (15.0–36.0)	23.0 (12.0–35.0)	20.0 (9.0–33.0)	<.001
Pre-pregnancy diabetes	3 (0.3)	65 (0.4)	173 (1.1)	288 (2.3)	191 (2.9)	167 (3.6)	<.001
Gestational diabetes	30 (2.9)	594 (4.0)	984 (6.0)	1,310 (10.5)	894 (13.5)	740 (16.1)	<.001
Pre-pregnancy hypertension	5 (0.5)	71 (0.5)	163 (1.0)	221 (1.8)	164 (2.5)	222 (4.8)	<.001
Gestational hypertension	30 (2.9)	442 (3.0)	666 (4.0)	802 (6.4)	608 (9.2)	534 (11.6)	<.001
Hypertension eclampsia	6 (0.6)	50 (0.3)	96 (0.6)	103 (0.8)	85 (1.3)	72 (1.6)	<.001
Previous preterm birth	34 (3.3)	503 (3.4)	587 (3.6)	568 (4.6)	314 (4.8)	248 (5.4)	<.001
Payment
Medicaid	546 (53.0)	7,779 (52.6)	8,929 (54.3)	7,219 (57.8)	3,821 (57.8)	2,727 (59.2)
Private Insurance	269 (26.1)	4,120 (27.9)	4,291 (26.1)	3,315 (26.5)	1,857 (28.1)	1,322 (28.7)
Self-pay	103 (10.0)	1,323 (9.0)	1,513 (9.2)	801 (6.4)	310 (4.7)	191 (4.2)
Other	112 (10.9)	1,565 (10.6)	1,722 (10.5)	1,156 (9.3)	621 (9.4)	363 (7.9)	<.001
Other birth outcomes
APGAR score‡
A score of 0–3	11 (1.1)	71 (0.5)	95 (0.6)	61 (0.5)	36 (0.6)	31 (0.7)
A score of 4–6	25 (2.4)	223 (1.5)	286 (1.7)	213 (1.7)	116 (1.8)	103 (2.2)
A score of 7–8	148 (14.5)	2,029 (13.8)	2,389 (14.5)	1,849 (14.9)	1,004 (15.2)	731 (15.9)
A score of 9–10	839 (82.0)	12,423 (84.3)	13,621 (82.8)	10,330 (83.0)	5,436 (82.5)	3,724 (81.2)	<.001
Final route and method of birth‡
Spontaneous	771 (74.9)	10,901 (73.7)	11,410 (69.3)	8,144 (65.2)	4,078 (61.7)	2,468 (53.6)
Forceps	8 (0.8)	129 (0.9)	136 (0.8)	75 (0.6)	36 (0.5)	27 (0.6)
Vacuum	33 (3.2)	502 (3.4)	371 (2.3)	238 (1.9)	84 (1.3)	49 (1.1)
Cesarean	218 (21.2)	3,254 (22.0)	4,536 (27.6)	4,033 (32.3)	2,410 (36.5)	2,059 (44.7)	<.001

BMI, body mass index. GWG, gestational weight gain. IOM, Institute of Medicine.

^*P-value was calculated compared to the healthy weight BMI group.

^†Labelled characteristics are not normally distributed. Median with Q1-Q3 is expressed.

^‡Missing values for the labelled characteristics: marital status (n=3,830), APGAR score (n=181), final route and method of birth (n=5).

Table 2 shows both unadjusted and adjusted associations between pre-pregnancy BMI and PTB, between rate of GWG and PTB, and between rate of GWG and PTB by pre-pregnancy BMI categories among Pacific Islanders. In the adjusted analysis, compared to mothers with rate of GWG within the IOM guidelines, mothers with rate of GWG below the guidelines were more likely to experience PTB if their pre-pregnancy BMI indicated underweight (aHR=1.84, 95% CI 1.10–3.06), healthy weight (aHR=1.38, 95% CI 1.15–1.65), obesity I (aHR=1.22, 95% CI 0.97–1.52), or obesity II (aHR=1.43, 95% CI 1.05–1.96). Across all pre-pregnancy BMI groups, 186.3 excess PTB per 10,000 births were associated with the rate of GWG below the IOM guidelines compared with the rate of GWG within the guidelines (Table S3). Mothers with a rate of GWG above the guidelines did not have evidently different risk of PTB, except for mothers with underweight (aHR=1.57, 95% CI 0.92–2.69), overweight (aHR=0.88, 95% CI 0.75–1.02), or obesity II pre-pregnancy (aHR=1.31, 95% CI 0.98–1.76). Across all BMI groups, 16.3 excess PTB per 10,000 births were associated with rate of GWG above the IOM guidelines compared with the rate of GWG within the guidelines (Table S3). No association was observed between PTB and rate of GWG among mothers indicated pre-pregnancy obesity III (rate of GWG below the guidelines: aHR=1.01, 95% CI 0.73–1.41; rate of GWG above the guidelines: aHR=0.92, 95% CI 0.67–1.25).

Table 2.

Unadjusted and adjusted Hazard Rate Ratio Estimates for Pre-pregnancy BMI, Rate of GWG, and Preterm Birth among Pacific Islander Mother-Infant Dyads in the US, 2014–2018

Characteristics	Preterm birth (n=5,192) vs. term birth (n=50,783)
	Unadjusted models†			Adjusted model*,†
	HR	95% CI	P-value	HR	95% CI	P-value
Pre-pregnancy BMI
Underweight	1.50	(1.25–1.80)	<.001	-	-	-
Healthy weight	Ref.	Ref.	Ref.	-	-	-
Overweight	1.10	(1.02–1.18)	0.017	-	-	-
Obesity I	1.11	(1.03–1.21)	0.008	-	-	-
Obesity II	1.09	(0.99–1.21)	0.068	-	-	-
Obesity III	1.18	(1.06–1.31)	0.003	-	-	-
Rate of GWG (lb/week) group
Pre-pregnancy BMI: underweight
Below the IOM guidelines	1.94	(1.16–3.23)	0.0130	1.84	(1.10–3.06)	0.021
Within the IOM guidelines	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
Above the IOM guidelines	1.67	(0.98–2.86)	0.0590	1.57	(0.92–2.69)	0.101
Pre-pregnancy BMI: healthy weight
Below the IOM guidelines	1.44	(1.20–1.72)	<.001	1.38	(1.15–1.65)	<.001
Within the IOM guidelines	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
Above the IOM guidelines	1.01	(0.85–1.21)	0.884	0.99	(0.83–1.18)	0.880
Pre-pregnancy BMI: overweight
Below the IOM guidelines	1.02	(0.86–1.20)	0.874	1.05	(0.88–1.24)	0.632
Within the IOM guidelines	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
Above the IOM guidelines	0.83	(0.72–0.97)	0.519	0.88	(0.75–1.02)	0.088
Pre-pregnancy BMI: obesity I
Below the IOM guidelines	1.24	(0.99–1.56)	0.060	1.22	(0.97–1.52)	0.096
Within the IOM guidelines	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
Above the IOM guidelines	1.06	(0.87–1.32)	0.570	1.03	(0.83–1.26)	0.824
Pre-pregnancy BMI: obesity II
Below the IOM guidelines	1.35	(0.99–1.85)	0.060	1.43	(1.05–1.96)	0.027
Within the IOM guidelines	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
Above the IOM guidelines	1.35	(1.01–1.81)	0.045	1.31	(0.98–1.76)	0.073
Pre-pregnancy BMI: obesity III
Below the IOM guidelines	1.11	(0.80–1.54)	0.546	1.01	(0.73–1.41)	0.941
Within the IOM guidelines	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
Above the IOM guidelines	1.11	(0.81–1.52)	0.520	0.92	(0.67–1.25)	0.608

HR, hazard ratio, BMI, body mass index. GWG, gestational weight gain. IOM, Institute of Medicine.

^*Confounders included in the adjusted model include neonatal sex, maternal ethnicity, maternal age, mother’s nativity, mother’s education, death of prior children, parity>1, adequacy of prenatal care utilization index, enrolled in the WIC program, ever smoked, pre-pregnancy diabetes, gestational diabetes, pre-pregnancy hypertension, gestational hypertension, hypertension eclampsia, previous preterm birth, and health insurance payment (full table see Table S4).

^†Conclusions are based on both the magnitude of effect sizes (hazard ratios) and the coverage of 95% confidence intervals (CI). A wide 95%CI which covers approximately equally likely effects in either direction would be aligned with absence of an association. A 95%CI mostly covering effects in one direction and suggesting on average at least a 10% reduction, or conversely a 10% increase, in the rate of PTB aligns itself with a meaningful association.

The following four characteristics were identified to be associated with decreased risk of PTB (Table S4): maternal ethnicity (Guamanian and Samoan, vs Hawaiian), mother’s education (high school graduation or above vs under high school graduation), parity>1, and WIC enrollment. The following were associated with increased risk of PTB: male infant, maternal age (>=35 years vs 20–34 years), inadequate and adequate plus prenatal care vs adequate care, pre-pregnancy diabetes, GDM, pre-pregnancy hypertension, hypertension eclampsia, previous PTB, and mothers having self-paid or other insurance (vs Medicaid). Interactions between pre-pregnancy BMI and GDM, and between pre-pregnancy BMI and gestational hypertension did not show evidence of effect modification (Table S4). In the sensitivity analysis with gestational diabetes, gestational hypertension, and hypertension eclampsia excluded from the adjusted model, no evident changes were observed in the effect estimates of the retained variables (Table S5).

The predicted probability of PTB from the nested model (including pre-pregnancy BMI, rate of GWG, and their interaction only) is shown in Figure 2. In five of the pre-pregnancy BMI groups, mothers with a rate of GWG below the guidelines had a higher probability of experiencing PTB compared to mothers with a rate of GWG within the guidelines. In the remaining group – mothers with an overweight pre-pregnancy BMI – no difference was observed among mothers with GWG below or within the guidelines. There was a U-shaped association between rate of GWG above the guidelines and PTB across the pre-pregnancy BMI categories. In those with an underweight pre-pregnancy BMI, compared to mothers with a rate of GWG within the guidelines, mothers with a rate of GWG above the guidelines had a much higher probability of PTB. There were no observed differences within the healthy weight group, and the probability of PTB was lower among mothers with a rate of GWG above the guidelines in the overweight group. In all of the groups with pre-pregnancy obesity, mothers with a rate of GWG above the guidelines had a higher probability of PTB. Adjusting for other characteristics in the model (Figure 3), especially for mothers with a rate of GWG above the guidelines, illustrates how other identified risk factors can contribute to high probabilities of PTB at the individual level.

Figure 2.

The probability of preterm birth from the nested model distributed by pre-pregnancy BMI with the increment of gestational age among Pacific Islander mothers in the US, 2014–2018*

* Overlapped dashed lines were labelled with squares and zoomed in at the right above corners in corresponding panels: in panel Healthy Weight, “within the guideline” and “above the guideline” overlapped; in panel Overweight, “below the guideline” and “within the guideline” overlapped; in panel Obesity II, “below the guideline” and “above the guideline” overlapped; in panel Obesity III, “below the guideline” and “above the guideline” overlapped.

Figure 3.

Adjusted probability of preterm birth distributed by pre-pregnancy BMI with the increment of gestational age among Pacific Islander mothers in the US, 2014–2018*,†

* (A) Adjusted probability of preterm birth among mothers with rate of GWG below the guideline; (B) Adjusted probability of preterm birth among mothers with rate of GWG within the guideline; (C) Adjusted probability of preterm birth among mothers with rate of GWG above the guideline.

† Lower probability of PTB in panel (B) among pre-pregnancy underweight, obesity II, and obesity III may due to a small sample size.

Discussion

In this retrospective cohort study, among Pacific Islander mothers with a high cohort prevalence of obesity, associations between rate of GWG and PTB differed by pre-pregnancy BMI. Our findings highlight the importance of ongoing research among women with pre-pregnancy BMI in the highest obesity categories. Other identified risk factors for PTB were consistent with findings in other populations(23, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34), and are more likely to be observed among minority populations(35).

Total GWG below the IOM guidelines has been consistently associated with higher risk of PTB(11, 36). Low GWG can reflect micronutrient deficiencies, poor plasma volume expansion, diminished uteroplacental blood flow, and increased risk of inflammation; all important indicators for PTB(8). The mechanism linking excessive GWG and PTB is less clear; some authors hypothesize that weight gain-related inflammation is associated with early labor(8). There is, however, inconsistency among studies that examine associations between PTB and the interaction between pre-pregnancy BMI and rate of GWG(14, 37). Our findings among underweight mothers are consistent with Huang et al.’s study among Chinese women(14) where there was a U-shaped association between rate of GWG and PTB – a rate of GWG either below or above the guidelines was associated with higher risk of PTB, but findings among mothers in other pre-pregnancy BMI groups varied.

Some recent studies suggest that women with pre-pregnancy obesity should be encouraged to gain less weight than the current IOM guidelines recommend and others argue that no weight gain, or even some weight loss is optimal(38). Overall, our findings demonstrate that, among this Pacific Islander cohort, a rate of GWG below the guidelines is associated with increased risk of PTB, including among women with pre-pregnancy obesity (excluding pre-pregnancy obesity class III), which replicates findings of a recent multi-ethnic meta-analysis(12) and is indirectly supported by studies for other perinatal outcomes among women with pre-pregnancy obesity(39, 40). Our data show that a rate of GWG above the guidelines, may not be harmful for all Pacific Islander women. However, GWG above the guidelines may increase the risk of other maternal complications like GDM and gestational hypertension(41, 42), which in turn can influence the risk of PTB. This was evident when we plotted the probability of PTB from the nested (Figure 2) and the adjusted (Figure 3) model, where we noted the heterogeneity in the individual risk of PTB among women within the same pre-pregnancy BMI stratum, indicating the necessity of individual-specific GWG counseling, that considers each individual’s constellation of risk factors.

Our study included a greater proportion of women with pre-pregnancy BMI in the Obesity III range than other cohorts addressing similar research questions. Among mothers in the pre-pregnancy class III obesity group (8.2% of the cohort), we did not find rate of GWG either below or above the guidelines to be associated with PTB. In a study from the UK, extreme obesity (BMI >40; <0.1% of the study population) was reportedly associated with more complications compared to other pre-pregnancy BMI groups(43). Current knowledge, including our study, is likely to be limited by the fact that fewer pregnancies among women with severe obesity result in live births. In 2020, Pacific Islander mothers had the highest fetal mortality rate (10.6 per 1,000 live births and fetal deaths) which is approximately 1.4 times the rate among White (7.8)(44). In another study restricted to live-born singletons, the absence of association was also observed between small-for gestational-age and pre-pregnancy obesity class III(45). Therefore, future research should incorporate miscarriages or stillbirths in the sample of pregnant mothers to investigate whether the absence of association is due to a potential selection bias in live-birth datasets.

Limitations in this study should be noted. First, we identified differential risk of PTB by ethnicity subgroups but, with approximately 60% of included mothers classified as “Other Pacific Islanders”, we could not further explore this finding. Second, trimester-specific weight gain was not recorded, so we were limited to the assumption that all women gained equally in the first trimester and at an equal rate throughout pregnancy. This is a limitation of most administrative or surveillance data. Third, our knowledge of the risks associated with severe pre-pregnancy obesity is still limited and fetal death-related research may be needed to understand why no association between GWG and PTB was observed in this group. Fourth, records with missing values for pre-pregnancy BMI or GWG had poorer socio-economic characteristics, and even with imputation, our findings should be cautiously generalized. Finally, the reliability and measurement validity of included variables are based on the rigor of the data collection of the US NCHS(15). Despite these limitations, we used a national level live-birth dataset with the largest coverage of Pacific Islanders in the US who have been underrepresented in perinatal research and have a relative higher prevalence of severe pre-pregnancy obesity; GA was also measured by obstetric estimate which is more accurate than last menstrual period(23) in determining PTB.

Conclusion

This work adds to a paucity of perinatal literature focused on the health and birth outcomes of Pacific Islander women. As has been observed in other studies with similar questions, the relationships between pre-pregnancy weight status, gestational weight gain, and preterm birth are complex and clearly modified by other risk factors. Continued efforts to better identify those at risk for preterm birth are warranted given the high prevalence among this community.

What is already known about this subject?

• Both pre-pregnancy body mass index (BMI) and gestational weight gain are important risk factors for preterm birth. However, the association between pre-pregnancy BMI and PTB, and the association between rate of GWG and PTB modified by pre-pregnancy BMI varies by population.

What are the new findings in your manuscript?

• Among Pacific Islanders in the US, the association between rate of gestational weight gain below or above existing guidelines and preterm birth differs by pre-pregnancy BMI.

• No association was observed between rate of GWG and PTB in women with pre-pregnancy class III obesity; more data is needed to draw adequate conclusions for this group.

How might your results change the direction of research or the focus of clinical practice?

• This work highlights the need for additional, population-specific, investigation into associations among gestational weight gain, pre-pregnancy BMI and preterm birth.

• Individual risk factors contribute to the heterogeneity in risk for preterm birth among mothers with different combinations of pre-pregnancy BMI and gestational weight gain, necessitating future research in this area to inform weight gain counseling in clinical practice.