Women’s prepregnancy underweight as a risk factor for preterm birth: a retrospective study

Abstract

Objective

To investigate the distribution of known factors for preterm birth (PTB) by severity of maternal underweight; to investigate the risk adjusted relationship between severity of underweight and PTB and to assess if the relationship differed by gestational age.

Design

Retrospective cohort study.

Setting

State of California, USA.

Methods

Maternally linked hospital and birth certificate records of 950,356 California deliveries in 2007–2010 were analyzed. Singleton live births of women whose pre-pregnancy body mass index (BMI) was underweight (<18.5 kg/m2) or normal (18.50–24.99 kg/m2) were analyzed. Underweight BMI was further categorized as: severe (<16.00), moderate (16.00–16.99) or mild (17.00–18.49). PTB was grouped as 22–27, 28–31, 32–36 or <37 weeks (compared with 37–41 weeks). Adjusted multivariable Poisson regression modeling was used to estimate relative risk for PTB.

Main outcome measures

Risk of PTB.

Results

72,686 (7.6%) women were underweight. Increasing severity of underweight was associated with increasing percent PTB: 7.8% (n=4421) in mild, 9.0% (n=1001) in moderate and 10.2% (475) in severe underweight. The adjusted relative risk of PTB also significantly increased: aRR=1.22 (95%CI: 1.19–1.26) in mild, aRR=1.41 (95%CI: 1.32–1.50) in moderate and aRR=1.61 (95%CI: 1.47–1.76) in severe underweight. These findings were similar in spontaneous PTB, medically indicated PTB, and the gestational age groupings.

Conclusion

Increasing severity of maternal pre-pregnancy underweight BMI was associated with increasing risk adjusted PTB at <37 weeks. This increasing risk was of similar magnitude in spontaneous and medically indicated births and in preterm delivery at 28–31 and at 32–36 weeks of gestation.

Introduction

With approximately 13 million babies born each year before 37 weeks of gestation,¹ preterm birth (PTB) is a leading cause of infant mortality and neonatal morbidity.² Although multiple risk factors have been related to PTB, it continues to be a complex phenomenon without a cure.

One of the potentially modifiable risk factors for PTB is maternal body mass index (BMI).^{3, 4, 5} Both low (<18.5) and high (>29) BMI are shown to associate with PTB.^{4, 5} Recent studies have demonstrated that the relationship between obesity and prematurity is influenced by the extent of obesity, type of PTB, presence or absence of comorbidities, parity, and gestational age.^{4, 6, 7} In addition to the worldwide concern on the negative health effects of increasing obesity, maternal underweight and malnutrition are serious problems^8,9 that may have both short and long-term consequences.^10,11However, so far, only a few studies have investigated effects of underweight severity (severe, BMI <16; moderate, BMI 16 – 16.9; mild, BMI 17 – 18.49 kg/m²) on broad categories of gestational age reflecting PTB¹² and no large studies have addressed which factors are important to the relationship between underweight and PTB.

In this large population based cohort study, we had 4 specific aims (1) to assess the distribution of known risk factors for PTB by severity of maternal underweight, (2) to assess if the relationship between severity of underweight and PTB persisted after adjusting for these risk factors (3) to assess if the adjusted relationship between underweight severity and PTB differed by gestational age, and finally, (4) to examine the hypothesis that risk factors for PTB would have a larger effect size in underweight women compared to normal weight women.

Methods

Data for this study come from 2007–2010 California birth cohorts reflecting California vital statistics birth records linked with Office of Statewide Health and Planning (OSHPD) maternal and infant hospital discharge data. These data contain information on a range of maternal and pregnancy characteristics found on the birth certificate paired with clinical details from the delivery hospitalization for nearly all inpatient live births, and has been well described elsewhere.¹³ Stanford University Institutional Review Board and the California State Committee for the Protection of Human Subjects reviewed and approved this study.

Demographic risk factors for PTB derived from birth certificates included maternal race/ethnicity, age, height, prepregnancy weight, education, parity, receipt of prenatal care, payer for the delivery, and gestational age at delivery based on obstetric estimate reported on the birth certificate.¹⁴ In addition to maternal demographics we also included other potential behavioral risk factors for preterm birth. Maternal eating disorder was defined based on the International Classification of Diseases, 9^th revision, Clinical Modification (ICD-9-CM) codes (307.1, 307.50, 307.51) in hospital discharge data, as were maternal alcohol (303, 305.0) and drug (304, 305.2-.9, 648.3) use during pregnancy. Smoking was defined as either birth certificate information of ≥1 cigarettes per day during pregnancy or based on smoking related ICD-9 codes (305.1, 649.0) in the delivery hospital discharge data.

Maternal comorbidities that have been associated with preterm birth were identified from ICD9-CM codes reported on the birth hospitalization dataset. Specifically, comorbidities included in analysis were: pre-existing diabetes (ICD-9 CM 250, 648.0); gestational diabetes (648.8); pre-existing hypertension (401–405, 642.0, 642.1, 642.2, 642.7, 642.9); gestational hypertension/ preeclampsia/eclampsia (642.3, 642.4, 642.5, 642.6, 642.7); placenta previa (641.0, 641.1); placental abruption (641.2) and anemia (280–285, 648.2).

PTB was defined as a live birth occurring at less than 37 weeks of gestation but was also assessed in gestational categories of 22–27 weeks, 28–31 weeks, and 32–36 weeks, each compared to term births at 37–41 weeks. PTB was further subtyped based on maternal ICD-9-CM diagnosis and procedure codes along with birth certificate codes in a hierarchical classification.⁴ First, spontaneous PTBs were identified as those births <37 weeks with preterm premature rupture of membranes, premature labor, or tocolytics. Medically indicated PTBs were those induced or delivered by cesarean section <37 weeks and not previously identified as spontaneous. All births <37 weeks not captured by either above groups were considered unclassifiable.

Body mass index (BMI) was calculated based on recorded height and prepregnancy weight (BMI = weight (in kilograms) / height ² (in meters)). Underweight BMI was defined as <18.5 kg/m² and further categorized into severe (<16 kg/m²), moderate (16–16.9 kg/m²) and mild (17–18.49 kg/m²). Normal BMI was defined as 18.5–24.9 kg/m².¹⁵ Data on maternal prepregnancy weight and height were self-reported.

During 2007–2010, there were 2,027,983 singleton live birth vital statistic records linked with maternal and infant hospital discharge summaries. Of these, we identified births with the following primary exclusion criteria (not mutually exclusive): gestational age <22 or >41 weeks (n=30,617), missing maternal height (n=105,635), missing prepregnancy weight (n=158,395), and overweight or obese pregnancy BMI (n=844,893). This left a total of 985,773 births to underweight and normal BMI mothers of which we further excluded (not mutually exclusive) missing education (n=32,953), maternal age <13 or >55 years (n=26), missing parity or >10 (n=423), or missing race/ethnicity (n=16,297). Our final analytic cohort consisted of 950,356 singleton live births born between 22–41 weeks of gestation to normal or underweight women with complete covariate information. Because the purpose of the study was to assess the impact of underweight on PTB the cohort was limited to deliveries at 22–41 weeks gestation.

Statistical analysis

Statistical analysis was performed using SAS version 9.3 (SAS Institute, Cary, NC, USA). First we compared maternal and delivery variables (Table 1), maternal behavioral factors (Table 2) and medical conditions (Table 2) between underweight categories and women with normal pre-pregnancy BMI. Bivariate comparisons of continuous variables were performed using ANOVA, and of categorical variables using Chi-square test. The level of significance was set at p<0.05.

Table 1.

Characteristics of the study subjects with underweight pre-pregnancy BMI (severe, moderate, mild) and normal pre-pregnancy BMI.

		Underweight BMI categories (kg/m2)
Variable		Severe (<16) N=4665	Moderate (16–16.9) N=11108	Mild (17–18.49) N=56913	Normal (18.5–24.9) N=877670	P-value*
Maternal age in years (SD)		25.2 (6.1)	25.8 (6.2)	26.9 (6.5)	28.1 (6.4)	<0.001
Maternal height in inches (SD)		66.0 (4.1)	64.7 (3.1)	64.2 (2.8)	63.7 (2.7)	<0.001
Maternal						<0.001
race/ethnicity:
Non-Hispanic White		21% (968)	26% (2882)	30% (16919)	32% (279545)
Non-Hispanic Black		6% (265)	6% (706)	5% (3050)	5% (41730)
Asian		24%(1100)	24% (2693)	26% (14543)	14% (118945)
Hispanic		45%(2118)	38% (4268)	35% (19908)	45% (398216)
Other		5% (214)	5% (559)	4% (2493)	4% (39234)
Maternal education:						<0.001
Some high school or less		31% (1459)	25% (2725)	21% (11956)	22% (195074)
High school graduate		29% (1375)	29% (3224)	25% (14387)	24% (210910)
Some college		20% (946)	22% (2484)	22% (12347)	22% (191438)
College graduate or more		19% (885)	24% (2675)	32% (18223)	32% (280248)
Parity:						<0.001
1		58% (2691)	56% (6237)	55% (31169)	46% (402570)
≥2		42% (1974)	44% (4871)	45% (25744)	54% (475100)
Prenatal care initiation:						<0.001
In first 5 months		91% (4265)	92% (10194)	93% (53100)	94% (825314)
6 months or later/no initiation/unknown		9% (400)	8% (914)	7% (3813)	6% (52356)
Payer for prenatal care:
MediCal (Public)		59% (2738)	50% (5598)	44% (24790)	42% (367834)	<0.001
Private		34% (1590)	43% (4789)	51% (28754)	53% (464187)
NA/Uninsured/Unknown		4% (192)	3% (367)	3% (1672)	2% (19493)
Other		3% (145)	3% (354)	3% (1697)	3% (26156)
Prior preterm delivery at <37 weeks:						0.285
No		99% (4637)	99% (11034)	99% (56,602)	99% (872956)
Yes		1% (28)	1% (74)	1% (311)	1% (4714)
Preterm delivery (<37 weeks):						<0.001
Spontaneous		8% (350)	7% (734)	6% (3264)	5% (42670)
Medically indicated		2% (84)	2% (176)	1% (724)	1% (10283)
Unclassified		1% (41)	1% (91)	1% (433)	1% (5191)

BMI, body mass index

Data are presented as mean (SD) for maternal age and height, and % (N) for all the other variables

^*ANOVA or Chi-square test used for comparison.

Table 2.

Maternal behavioral factors and medical conditions in women with underweight pre-pregnancy BMI and normal BMI.

	Underweight BMI categories (kg/m2)
	Severe (<16) N=4665	Moderate (16–16.9) N=11108	Mild (17–18.49) N=56913	Normal (18.5–24.9) N=877670	P-value*
Maternal Behaviors
Smoking during pregnancy	5.40% (252)	5.30% (589)	4.12% (2346)	2.77% (24297)	<0.001
Alcohol use during pregnancy	0.11% (5)	0.09% (10)	0.11% (65)	0.10% (906)	0.84
Drug abuse during pregnancy	1.78% (83)	1.68% (187)	1.31% (745)	0.99% (8717)	<0.001
Eating disorder	0.06% (3)	0.06% (7)	0.02% (11)	0.01% (69)	<0.001
Maternal Medical Conditions
Anemia	9.65% (450)	9.48% (1053)	8.50% (4839)	7.93% (69641)	<0.001
Pre-existing diabetes	0.09% (4)	0.18% (20)	0.15% (87)	0.34% (2997)	<0.001
Gestational diabetes	3.52% (164)	3.38% (375)	3.17% (1804)	4.40% 38593	<0.001
Pre-existing hypertension	0.43% (20)	0.45% (50)	0.41% (236)	0.76% (6714)	<0.001
Gestational hypertension/Preeclampsia/Eclampsia	3.99% (186)	3.15% (350)	2.94% (1674)	3.78% (33215)	<0.001
Placental abruption	1.22% (57)	0.98% (109)	1.06% (601)	0.86% (7573)	<0.001
Placenta previa	0.73% (34)	0.64% (71)	0.73% (415)	0.68% (5981)	0.52

Data are presented as %(N);

^*Chi-square test used for comparison

The association between underweight BMI categories and PTB was measured with adjusted relative risks (aRR) and 95% confidence intervals (CI) derived from multivariable Poisson regression models. Potential confounders were selected based on their significance in the univariable analysis (P<0.1) and because they have been associated with increased risk for PTB. The potential confounders included in the multivariable model were maternal age (continuous), prenatal care initiation (during first 5 months, 6 months or later/no initiation/unknown), maternal education (some high school or less, high school graduate, some college, college graduate or more), race/ethnicity (Non-Hispanic White, Non-Hispanic Black, Asian, Hispanic, Other), parity (nulliparous, multiparous), smoking during pregnancy (yes, no), presence of eating disorder (yes/no), anemia (yes/no), pre-existing diabetes (yes/no), pre-existing hypertension (yes/no), gestational diabetes (yes/no), gestational hypertension (yes/no), placental abruption (yes/no) and maternal height (continuous) to reduce further potential residual confounding associated with the BMI algorithm.¹⁶ Because risk adjusted PTB could result from medical intervention, we compared the relationship between severity of underweight and PTB in births that were spontaneous and in births that resulted from medical intervention.

Results

Among a total of 950,356 women included in analyses, 72,686 (7.6%) had an underweight pre-pregnancy BMI; 0.5% had severe, 1.2% moderate, and 6.0% mild underweight. An increasing trend of all PTBs (<37 weeks) was noted by increasing severity of maternal underweight: 7.8% in mild (n=4421 PTBs), 9.0% in moderate (n=1001) and 10.2% in severe underweight (n=475). The unadjusted relative risks were RR=1.17 (95%CI: 1.14–1.21) in mild, RR=1.36 (95%CI:1.28–1.45) in moderate and RR=1.54 (95%CI:1.40–1.68) in the severe underweight group.

Maternal demographics, medical conditions and behavioral risk factors for PTB and their relationship to severity of underweight are presented in Tables 1 and 2. Compared to women with a normal BMI, women with severe underweight were slightly younger, less likely to be non-Hispanic White, less educated more often nulliparous and received public (MediCal) insurance more often. The percentage of late prenatal care initiation was highest with severe underweight as were the percentages of spontaneous and medically indicated PTB deliveries. However, the rate of prior PTBs was similar across all underweight categories (Table 1).

Women with mild, moderate, and severe underweight demonstrated higher frequencies of smoking and drug abuse during pregnancy compared to normal weight women, whereas alcohol use during pregnancy was similar between the groups. Anemia during pregnancy was increased among those underweight compared to women of normal BMI, but underweight women were less likely to have pre-existing diabetes, gestational diabetes or pre-existing hypertension compared to women with normal BMI (Table 2).

Table 3 compares the observed relationship between severity of underweight and % PTB to the risk adjusted relationship for (1) all preterm births, for (2) the 73.4% of preterm deliveries that were spontaneous, and for (3) the 17.6% that were medically indicated (note: 9.0% could not be classified). The observed relative risk (RR) for PTB (< 37 weeks) increased from RR=1.17 (95%CI:1.14–1.21) in mildly underweight women to RR=1.54 (95%CI:1.40–1.68) in those who were severely underweight. Although the risk adjusted aRR’s were slightly higher than the observed RRs (mild: aRR=1.22; 95%CI:1.19–1.26 and severe: aRR=1.61; 95%CI:1.47–1.76), there were no statistically significant differences in the extent of the association between severity of underweight and increasing PTB as estimated by RR and aRR for all deliveries, spontaneous preterm deliveries and medically indicated deliveries. Furthermore, the extent to which increasing severity of underweight was associated with increasing risk for PTB, was similar for all preterm deliveries, spontaneous preterm deliveries, and medically indicated preterm deliveries (Table 3).

Table 3.

Observed unadjusted and risk adjusted associations between all preterm births (All PTB) and preterm birth subtypes (Spontaneous PTB and Medically indicated PTB) and maternal underweight categories versus normal BMI.

	All PTB <37 weeks			Spontaneous PTB <37 weeks			Medically indicated PTB <37 weeks
BMI (kg/m2)	N	RR (CI)	aRR (CI)	N	RR (CI)	aRR (CI)	N	RR (CI)	aRR (CI)
Severe underweight (<16)	475	1.54 (1.40,1.68)	1.61 (1.47,1.76)	350	1.56 (1.40,1.73)	1.62 (1.46,1.81)	84	1.59 (1.28,1.97)	1.78 (1.43,2.21)
Moderate underweight (16–16.99)	1001	1.36 (1.28,1.45)	1.41 (1.32,1.50)	734	1.37 (1.27,1.47)	1.41 (1.31,1.51)	176	1.38 (1.19,1.60)	1.53 (1.32,1.78)
Mild underweight (17–18.49)	4421	1.17 (1.14,1.21)	1.22 (1.19,1.26)	3264	1.18 (1.14,1.23)	1.22 (1.18,1.27)	724	1.10 (1.02,1.18)	1.24 (1.15,1.34)
Normal (18.5–24.99)	58144	ref (1.0)	ref (1.0)	42670	ref (1.0)	ref (1.0)	10283	ref (1.0)	ref (1.0)

RR, observed unadjusted relative risk

aRR, adjusted relative risk

BMI, body mass index

CI, confidence interval

Model adjustments include: maternal age, height, prenatal care initiation, maternal education, race/ethnicity, parity, smoking, drug abuse, presence of eating disorder, anemia, preexisting diabetes, preexisting hypertension, gestational diabetes, gestational hypertension and placental abruption.

Because the relationship between obesity and PTB has been shown to increase with decreasing gestational age groupings we examined the potential effect of gestational age in underweight women. The risk adjusted relationship between severity of underweight and PTB is shown for three gestational groupings in table 4. Although estimates at 22–27 weeks were limited by small sample size, based on the overlapping Cis, there was no evidence of an increase in effect size as estimated by aRRs in the moderately preterm (28–31 weeks) and late preterm (32–36 weeks) groupings (Table 4). Exclusion of cases with a history of prior PTB, did not significantly change the adjusted RRs across underweight categories for all PTBs, or the above groupings (Table 1S).

Table 4.

The association between underweight BMI categories and all preterm births (PTB) at 22–27 weeks, 28–31 weeks and 32–36 weeks of gestation compared to normal BMI presented as adjusted relative risks (aRR).

	PTB 22–27 weeks		PTB 28–31 weeks		PTB 32–36 weeks
BMI (kg/m2)	N	aRR (CI)	N	aRR (CI)	N	aRR (CI)
Severe underweight (<16)	16	1.22 (0.74,1.99)	36	1.62 (1.16,2.25)	423	1.65 (1.50,1.82)
Moderate underweight (16–16.99)	40	1.24 (0.90,1.69)	90	1.64 (1.33,2.03)	871	1.41 (1.31,1.50)
Mild underweight (17–18.49)	181	1.18 (1.01,1.37)	338	1.25 (1.12,1.40)	3902	1.23 (1.19,1.27)
Normal (18.5–24.99)	2342	ref (1.0)	4490	ref (1.0)	51312	ref (1.0)

BMI, body mass index

aRR, adjusted relative risk

CI, confidence interval

Model adjustments included: maternal age, height, prenatal care initiation, maternal education, race/ethnicity, parity, smoking, drug abuse, presence of eating disorder, anemia, preexisting diabetes, preexisting hypertension, gestational diabetes, gestational hypertension and placental abruption.

Table 2S examines the hypothesis that risk factors for PTB will have a larger effect size in underweight than in normal weight women. For each risk factor, the adjusted relative risk of the PTB at <37 weeks among underweight women (all categories) and among normal weight women are compared. While no association was noted in normal weight women, decreased parity was related to decreased risk of PTB among underweight women (Table 2S). Although limited by small numbers, eating disorder was associated with an increased risk for PTB among underweight women and not among normal weight women. The relative risk of preterm delivery associated with smoking, drug and alcohol use, anemia, pre-existing diabetes, pre-existing hypertension, gestational hypertension, placenta previa and placental abruption were similar in both underweight and normal weight women (Table 2S).

Discussion

Main Findings

Our results based on almost a million live births demonstrate that the risk for PTB increased with the severity of underweight, and that this relationship persisted even after adjusting for maternal characteristics, pre-existing maternal comorbidities and behavioral risk factors. We also found that the strength of the relationship was similar in medically indicated and spontaneous births, and at 28–31 and 32–36 weeks of gestation. Although we hypothesized that risk factors for PTB would have a greater effect in underweight women, our results did not support this hypothesis.

Strengths and limitations

The population based California data in this study allowed for PTB stratification by three gestational age groups as well as by clinical subtypes and increases the generalizability of our findings. A limitation is that our data were derived from birth certificates and discharge databases with their inherent errors. Some behavioral factors that were investigated, i.e. drug abuse and eating disorder, have been shown to be poorly recorded^{16, 17} thus, could have been underreported in this study. In addition, we were limited to investigate BMI derived from self-reported weight and height information, which has previously shown to relate to biased risk estimates of PTB.¹⁸ Lastly, this large-scale study cannot identify specific mechanisms underlying the association between underweight and PTB. Our study, however, adds to the literature and, importantly, may offer background for more specific, mechanistic studies.

Interpretation

Our results confirm the findings of prior studies among underweight women^{5, 12} and bring new insights to this rather understudied topic. Previously, some smaller, single center studies have documented the relationship between underweight and preterm delivery^19,20, however, without categorizing underweight by its severity. In addition, Salihu et al. showed that women in all underweight categories had increased risk of PTB at <37 weeks and at <33 weeks and that the extent of risk was dependent upon the severity of underweight.¹² With our substantially larger cohort, we were able to demonstrate a relationship between severity of underweight and PTB at both 28–31 and 32–36 weeks, even after multiple adjustments for confounders. In addition, after excluding women with prior preterm birth, we found that the relationship between underweight categories and the risk of PTB at 28–31; 32–36 and <37 weeks of gestation remained essentially unchanged. Furthermore, although at 22–27 weeks gestation we were likely underpowered to adequately investigate differences among severe and moderate underweight groups, the risk of PTB was significantly higher among mild underweight (n=181) compared to normal weight women (n=2342).

Previously, multiple studies have demonstrated the complex relationship between maternal obesity and increased risk of PTB.²¹ While both obesity and underweight increase the risk of PTB with increasing severity^{7, 12}, many of the factors that play a critical role in the obesity-preterm relationship^{4, 6, 7} do not affect the relationship between underweight and PTB. For example, in this study, risk of PTB was similar in medically indicated and spontaneous deliveries.

Although several factors have been proposed, the mechanisms behind a PTB of underweight women are unknown. We investigated the effect of maternal pre-existing conditions and behavioral factors on PTB. Not surprisingly, women in all underweight categories had less diabetic and hypertensive disorders compared to normal weight women, whereas anemia occurred more often in underweight women. However, the independent relative risks of PTB associated with smoking, drug and alcohol use, anemia, pre-existing diabetes, pre-existing hypertension, gestational hypertension, placenta previa and placental abruption were similar in underweight and normal weight women. Thus, although behavioral and medical risk factors for PTB among underweight women are of concern, we found insufficient evidence that their effect size was enhanced in underweight compared to normal weight women. Although limited by a small number of cases, maternal eating disorder was associated with PTB among underweight women and not in normal weight women, which is in line with the recent study by Linna et al.²² It is possible that the association between maternal underweight and PTB exists directly due to the lack of nutrients or there is an indirect subtle effect of multiple other behavioral factors like smoking, poor diet and medical illness. Studies with more detailed information on such factors are needed to investigate the complex relationship between underweight and PTB.

Conclusion

In conclusion, our study points out that women in all underweight BMI categories at prepregnancy are at increased risk of preterm delivery, even after adjustment for maternal characteristics, comorbidities and behavioral factors related to maternal underweight. Based on our California estimate that 7.6% of pregnant women are underweight, our findings support the potential importance of interventions to reduce prepregnancy underweight as an important strategy to reduce premature births.