HIV Infection and Its Impact on Fetal Outcomes Among Women of Advanced Maternal Age: A Propensity Score Weighted Matching Approach

Abstract

Advanced maternal age (AMA) and HIV status have been investigated separately for their influence on infant outcomes. Both are associated with adverse fetal growth outcomes, including low birth weight (LBW) and preterm birth (PTB). However, the impact of the cooccurrence of these factors in relation to birth outcomes remains relatively understudied. We analyzed Florida hospital discharge data linked to vital records. The study population consisted of women who had a singleton live birth between 1998 and 2007 (N=1,687,176). The exposure variables were HIV infection and maternal age, while the outcomes of interest were LBW, PTB, and small for gestational age (SGA). We matched HIV-positive women to HIV-negative women on selected variables using propensity scores. To approximate relative risks, we computed adjusted odds ratios (AOR) and 95% confidence intervals (CI) generated from logistic regression models and accounted for the matched design using the generalized estimating equations framework. After adjusting for demographic variables, clinical conditions, and route of birth, the risks of LBW, PTB, and SGA remained significant for HIV-positive women, regardless of age. HIV-positive women of AMA (≥35 years) were more likely to have infants of LBW (AOR=1.73, 95% CI=1.37–2.18), PTB (AOR=1.35, 95% CI: 1.06–1.71), and SGA (AOR=1.52, 95% CI=1.22–1.89), compared to uninfected mothers of younger age (<35 years). For women of advanced age, HIV positivity elevates their risk for LBW and PTB. The interplay of HIV status and age should be considered by healthcare providers when determining appropriate interconception strategies for women and their families.

Introduction

A dvanced maternal age (AMA) is defined as a mother who is aged 35 years or older at the time of delivery.¹ AMA is a well-known risk factor for poor birth outcomes.^1

–7 Older women have a higher likelihood of experiencing a preterm birth (PTB) and having an infant of low birth weight (LBW).^2
–4 Advanced maternal age is also a risk factor for intrauterine growth retardation (IUGR).⁷ These adverse outcomes have been noted to disproportionately affect African-American and Hispanic women of AMA.⁵

HIV seropositivity has also been implicated in feto-infant morbidities. Studies have identified an association between maternal HIV status and LBW,^8
–10 PTB,^9
–11 small for gestational age (SGA),¹⁰ and IUGR.⁸ The severity of maternal HIV disease, as indicated by CD4 counts, shows a strong association with fetal growth restriction.¹² Additionally, maternal age has been found to be a risk factor for fetal growth restriction among HIV-positive women.¹²

Given that maternal age and HIV are both associated with fetal growth restriction, it is likely that older pregnant women with HIV are at increased risk for LBW and PTB; however, this association has been underresearched, with few studies focused specifically on the interaction between advanced maternal age and HIV status. Only one known study by Brown and colleagues revealed an association between PTB and HIV infection in older mothers; however, this study sample consisted of only 142 women.¹³ The current study was designed to further investigate this phenomenon with a much larger population-based dataset.

Materials and Methods

We conducted a retrospective population-based cohort study of women having a singleton live birth in the state of Florida between 1998 and 2007 (inclusive). The Hospital Inpatient Discharge (HID) data linked with vital records data for the state of Florida were utilized for this study. HID data were obtained from the Florida Agency for Health Care Administration (AHCA), and the vital records were acquired from the Florida Department of Health (FDOH). Prior to initiation, this study was approved by the Institutional Review Board of the University of South Florida. The deidentified linked data consist of a total of 1,700,734 births to women in Florida within the duration of the study. After eliminating women with missing data and whose gestational age was outside the range of viability (20–44 weeks of gestation), 1,687,176 births were retained for analyses. We used ICD-9 codes in the HID data to identify maternal HIV infection status (042, 07953, V08).

The primary outcomes of interest in this study were LBW (birth weight <2,500 g), PTB (<37 gestational weeks), and small for gestational age (SGA: infants weighing below the tenth percentile of birth weight for their gestational age using normalized growth curves).¹⁴ Gestational age was computed in weeks by calculating the number of weeks between the first day of the last menstrual period and the date of delivery of the fetus. When the menstrual estimate of gestational age was inconsistent with the birth weight (e.g., very low birth weight at term), a clinical estimate of gestational age on the vital records was used instead.¹⁵

All demographic variables, including race/ethnicity (white non-Hispanic, black non-Hispanic, white Hispanic, black Hispanic, or other), maternal age (<35 or ≥35 years old at the time of delivery), marital status (unmarried or married), education (<12 or ≥12 years), parity (nulliparous or multiparous), cigarette smoking during pregnancy (yes or no), and adequacy of prenatal care (adequate or inadequate), were extracted from birth certificate data. Adequacy of prenatal care utilization was measured using the revised graduated index algorithm (R-GINDEX).¹⁶ The R-GINDEX has been found to be more accurate than several others¹⁷ in describing the level of prenatal care utilization among groups that are considered to be high risk.¹⁸ The R-GINDEX assesses the adequacy of care based on the trimester prenatal care began, the number of visits, and the gestational age of the infant at birth.¹⁶ Alcohol abuse and drug abuse were determined using ICD-9 codes. Alcohol abuse was defined by codes that indicate acute alcohol intoxication (303.00-03), alcohol dependence (303.90-93), nondependent alcohol abuse, such as “binge drinking” (305.00-03), alcohol-induced mental disorders (291.0-5, 9; 291.81-82, 89), and alcohol affecting the fetus or newborn via placenta or breast milk (760.71). ICD-9 codes signifying drug dependence complicating pregnancy, childbirth, or the puerperium or postdelivery period (648.30-648.34) were utilized to determine drug abuse. Tobacco use during pregnancy was abstracted from both hospital inpatient discharge data and vital records. Within the HID data, ICD-9 codes that denote tobacco use disorder or excessive tobacco use that is harmful to a person's health or social functioning (305.10), complicates pregnancy, childbirth, or puerperium (649.00-04), or has toxic effects (989.84) were utilized in this study. Pregnancy complications included in this study were obtained from the hospital discharge data and were based on ICD-9 principal and other diagnostic codes, as follows: anemia (280, 2800, 2801, 2808, 2809, 2810, 2811, 2812, 2813, 2818, 2819, 2820, 2821, 2822, 2823, 2824, 28241, 28242, 28249, 2825, 2826, 28260, 28261, 28262, 28263, 28264, 28268, 28269, 2827,2828, 2829, 2830, 28310, 28311, 28319, 2832, 2839, 2840, 28401, 28409, 2841, 2842, 2848, 28481, 28489, 2849, 2850, 2851, 2852, 28521, 28522, 28529, 2853, 2858, 648.2); gestational diabetes (648.8); diabetes mellitus (250, 648.0); gestational hypertension (642.3); chronic hypertension (642.0, 401.0, 401.1, 401.9, 642.1, 642.2, 742.7); preeclampsia (642.4, 642.5, 642.7, 642.9); eclampsia (642.6); placental abruption (641.2); and placental previa [maternal (641.0, 641.1), infant (762.0)].

Statistical analysis

We compared baseline characteristics (i.e., demographic characteristics and pregnancy complications) between mothers who were HIV positive versus those who were HIV negative using McNemar's Chi-square test for matched categorical data. Because the baseline characteristics of HIV-positive mothers (cases) differed from those of HIV-negative mothers (controls), we applied weighted propensity score methods to adjust for these differences.¹⁹ Observational studies, such as ours, are nonrandomized, and the significant differences in demographic and clinical baseline characteristics (presented in Tables 1 and 2) were not unexpected. The propensity score technique allows us to balance for observed covariates that might potentially influence group assignment and determine the effect of exposure by weighting both the cases and controls with a single composite measure computed from all covariates of interest.

Table 1.

Demographic Characteristics of the Study Population by Maternal HIV Status Before and After Propensity Weighting (Florida, 1998–2007)

	Before propensity weighting			After propensity weighting
Characteristics	HIV negative N=1,682,542 n (%)	HIV positive N=4634 n (%)	p-value^a	HIV negative N=9,268 n (%)	HIV positive N=4,634 n (%)	p-value^a
Maternal age			<0.0001			0.63
<35 years	241,954 (14.38)	554 (11.96)		8,134 (87.76)	4,080 (88.04)
≥35 years	1,440,588 (85.62)	4,080 (88.04)		1,134 (12.24)	554 (11.96)
Race/ethnicity			<0.0001			0.75
White non-Hispanic	900,587 (53.53)	597 (12.88)		1,182 (12.75)	597 (12.88)
Black non-Hispanic	355,636 (21.14)	3,441 (74.26)		6,884 (74.28)	3,441 (74.26)
White Hispanic	287,098 (17.06)	247 (5.33)		465 (5.02)	247 (5.33)
Black Hispanic	28,646 (1.70)	240 (5.18)		489 (5.28)	240 (5.18)
Other	110,575 (6.57)	109 (2.35)		248 (2.68)	109 (2.35)
Marital status			<0.0001			0.98
Unmarried	660,153 (39.24)	3,530 (76.18)		7,062 (76.20)	3,530 (76.18)
Married	1,022,389 (60.76)	1,104 (23.82)		2,206 (23.80)	1,104 (23.82)
Education			<0.0001			0.71
≤12 years	299,405 (17.76)	1,836 (39.62)		3,642 (39.30)	1,836 (39.62)
≥12 years	1,383,137 (82.21)	2,798 (60.38)		5,626 (60.70)	2,798 (60.38)
Parity			<0.0001			0.84
Nulliparous	579,327 (34.43)	950 (20.50)		1,914 (20.65)	950 (20.50)
Multiparous	1,103,215 (65.57)	3,684 (79.50)		7,354 (79.35)	3,684 (79.50)
Prenatal care			<0.0001			0.87
Inadequate	797,859 (47.42)	2,917 (62.95)		5,821 (62.81)	2,917 (62.95)
Adequate	884,683 (52.58)	1,717 (37.05)		3,447 (37.19)	1717 (37.05)
Tobacco use			<0.0001			0.82
No	1,517,788 (90.21)	3,988 (86.06)		7,989 (86.20)	3,988 (86.06)
Yes	164,754 (9.79)	646 (13.94)		1,279 (13.80)	646 (13.94)
Alcohol use			<0.0001			0.86
No	1,675,010 (99.55)	4,554 (98.27)		9,104 (98.23)	4,554 (98.27)
Yes	7,532 (0.45)	80 (1.73)		164 (1.77)	80 (1.73)
Drug abuse			<0.0001			0.92
No	1,668,341 (99.16)	4,370 (94.30)		8,736 (94.30)	4,370 (94.30)
Yes	14,201 (0.84)	264 (5.70)		532 (5.70)	264 (5.70)

Significant values are in bold font.

p-values<0.05 are considered significant.

Table 2.

Selected Pregnancy Complications of Study Population by Maternal HIV Status Before and After Propensity Weighting (Florida, 1998–2007)

	Before propensity weighting			After propensity weighting
Pregnancy complications	HIV negative N=1,682,542 n (%)	HIV positive N=4634 n (%)	p-value ^a	HIV negative N=9,268 n (%)	HIV positive N=4,634	p-value ^a
Anemia			<0.0001			0.84
No	1,563,097 (92.90)	3,959 (85.43)		7,930 (85.56)	3,959 (85.43)
Yes	119,445 (7.10)	675 (14.57)		1,338 (14.44)	675 (14.57)
Gestational hypertension			<0.0001			0.72
No	1,624,559 (96.55)	4,531 (97.78)		9,053 (97.68)	4,531 (97.78)
Yes	57,983 (3.45)	103 (2.22)		251 (2.32)	103 (2.22)
Existing hypertension			<0.0001			0.84
No	1,659,014 (98.60)	4,482 (96.72)		8,970 (96.78)	4,482 (96.72)
Yes	23,528 (1.40)	152 (3.28)		298 (3.22)	152 (3.28)
Gestational diabetes			<0.0001			0.88
No	1,606,881 (95.50)	4,506 (97.24)		9,016 (97.28)	4,506 (97.24)
Yes	75,661 (4.50)	128 (2.76)		252 (2.72)	128 (2.76)
Diabetes mellitus			0.02			0.81
No	1,670,521 (99.29)	4,587 (98.99)		9,178 (99.03)	4,587 (98.99)
Yes	12,021 (0.71)	47 (1.01)		90 (0.97)	47 (3.71)
Preeclampsia			0.05			0.31
No	1,628,662 (96.80)	4,462 (96.29)		8,891 (95.93)	4,462 (96.29)
Yes	53,880 (3.20)	172 (3.71)		377 (4.07)	172 (0.13)
Eclampsia			0.46			0.64
No	1,680,929 (99.90)	4,628 (99.87)		9,253 (99.84)	4,628 (99.87)
Yes	1,613 (0.10)	6 (0.13)		15 (0.16)	6 (0.13)
Placental abruption			0.64			0.33
No	1,666,964 (99.07)	4,588 (99.01)		9,159 (98.82)	4,588 (99.01)
Yes	15,578 (0.93)	46 (0.99)		109 (1.18)	46 (0.99)
Placenta previa			0.51			0.73
No	1,673,336 (99.45)	4,612 (99.53)		9,220 (99.53)	4,612 (99.53)
Yes	9,206 (0.55)	22 (0.47)		48 (0.47)	22 (0.47)
Placenta accreta			0.1			0.75
No	1,677,852 (99.72)	4,627 (99.85)		9,256 (99.87)	4,627 (99.85)
Yes	4,690 (0.28)	7 (0.15)		12 (0.13)	7 (0.15)
Route of delivery			<0.0001			1.00
Vaginal	1,192,118 (70.85)	2,051 (44.26)		4,102 (44.26)	2,051 (44.26)
Cesarean section	490,424 (29.15)	2,583 (55.74)		5,166 (55.74)	2,583 (55.74)

Significant values are in bold font.

p-values<0.05 are considered significant.

We used a multivariate logistic regression models to calculate the predicted probability of the dependent variable (i.e., probability of exposure to HIV infection) and the propensity score for each observation in the dataset (presented in Table 3). In model 1, we calculated the propensity score by weighting for demographic variables. In model 2, we calculated the propensity score by weighting for both demographic and pregnancy complications. Then, we used the propensity score to represent the relationship between all the covariates (presented in Tables 1 and 2) and our dependent variable (HIV infection status) by weighting each patient's data based on the inverse propensity of being in one of the two groups (HIV-positive or HIV-negative) when computing the estimated effect of exposure to HIV on birth outcomes.¹⁹ Each one of the HIV-positive mothers (cases) was matched to two HIV-negative mothers (controls) whose propensity scores were sufficiently close to that of the case. When there was more than one control that could be matched to a case, a match was selected at random from all identical controls using a random number with the RANUNI function in SAS (SAS Institute, Inc., Cary, NC, version 9.2). We opted to use one to two matching without replacement to maximize the precision of our estimates.

Table 3.

Adjusted Odds Ratios and 95% Confidence Intervals for the Association Between Maternal HIV Status, Age, and Feto-Infant Morbidity Outcomes (Florida, 1998–2007)

	Model 1 ^b,c			Model 2 ^b,d			Model 3 ^b,e
Maternal HIV status and age ^a	LBW AOR (95% CI)	PTB AOR (95% CI)	SGA AOR (95% CI)	LBW AOR (95% CI)	PTB AOR (95% CI)	SGA AOR (95% CI)	LBW AOR (95% CI)	PTB AOR (95% CI)	SGA AOR (95% CI)
HIV negative and young maternal age	Referent	Referent	Referent	Referent	Referent	Referent	Referent	Referent	Referent
HIV negative and advanced maternal age	1.20 (0.98–1.48)	1.14 (0.95–1.38)	0.98 (0.81–1.19)	1.37 (1.13–1.66)	1.28 (1.06–1.53)	1.00 (0.83–1.21)	1.17 (0.95–1. 43)	1.04 (0.85–1.27)	0.94 (0.78–1.15)
HIV positive and young maternal age	1.51 (1.35–1.70)	1.30 (1.17–1.45)	1.43 (1.29–1.58)	1.51 (1.35–1.69)	1.31 (1.18–1.46)	1.42 (1.28–1.57)	1.45 (1.30–1.62)	1.26 (1.13–1.40)	1.44 (1.30–1.59)
HIV positive and advanced maternal age	1.80 (1.42–2.28)	1.39 (1.10–1.77)	1.51 (1.21–1.88)	1.80 (1.42–2.28)	1.41 (1.11–1.79)	1.50 (1.20–1.87)	1.73 (1.37–2.18)	1.35 (1.06–1.71)	1. 52 (1.22–1.89)

Young maternal age: <35 years; advanced maternal age: ≥35 years.

Significant values are in bold font.

Model 1: Estimates were generated using GEE after propensity score weighting and matching of the exposed and unexposed groups with demographic variables (age, race, marital status, education, parity, adequacy of prenatal care, tobacco use, alcohol use, drug abuse).

Model 2: Estimates were generated using GEE after propensity score weighting and matching of the exposed and unexposed groups with demographic variables (Model 1) and pregnancy complications (anemia, gestational hypertension, existing hypertension, gestational diabetes, diabetes mellitus, preeclampsia, eclampsia, placental abruption, placenta previa, placenta accreta).

Model 3: Estimates were generated using GEE after propensity score weighting and matching of the exposed and unexposed groups with demographic variables (Model 1), pregnancy-related complications (Model 2), and route of delivery (vaginal vs. cesarean section).

LBW, low birth weight; PTB, preterm birth; SGA, small for gestational age; AOR, adjusted odds ratios; CI, confidence intervals.

Exposed and unexposed matched subjects are more similar than randomly selected exposed and unexposed subjects and, hence, do not form two independent samples.²⁰ Since the matching was done after the exposure, statistical analyses that do not account for the matched nature of the data will lead to a biased estimate of the outcome. To address bias in estimating our outcome, we used the Generalized Equation Estimation (GEE) method to estimate the risk for feto-infant morbidity outcomes (LBW, PTB, and SGA) among HIV-positive and negative mothers of advanced maternal age. GEE, which accounts for the paired nature of the data,²⁰ was conducted using the GENMOD procedure in SAS. The need to account for the matched design in the analysis of epidemiological studies is documented in the literature.²¹ All hypothesis tests were two-tailed with a type 1 error rate fixed at 5%.

Results

Maternal demographic characteristics and clinical conditions before and after propensity score weighted matching are presented in Table 1 and Table 2, respectively. Within the study population, a total of 4,634 mothers (0.27%) were HIV positive and 1,682,542 (99.73%) were HIV negative. Before propensity score weighted matching, demographic characteristics and pregnancy complications differed substantially between HIV-infected and uninfected groups. Compared to HIV-negative mothers, the HIV-positive mothers were more likely to be older, black (non-Hispanic), and unmarried women who used alcohol, drugs, and tobacco, had lower levels of education, and had previous pregnancies. Additionally, a higher proportion of HIV-infected mothers received inadequate prenatal care and experienced anemia, gestational hypertension, chronic hypertension, gestational diabetes, diabetes mellitus, and preeclampsia. After applying propensity score weighted matching, the two groups (HIV positive and HIV negative) became similar with respect to all covariates, as shown in Tables 1 and 2.

Compared to the referent group of HIV-negative young mothers (<35 years), HIV-positive mothers of advanced maternal age (≥35 years) demonstrated 41–80% increased odds of LBW [adjusted odds ratio (AOR)=1.73, 95% confidence interval (CI): 1.37–2.18], PTB (AOR=1.35, 95% CI: 1.06–1.71), and SGA (AOR=1.52, 95% CI: 1.22–1.89), after adjusting for selected demographic characteristics, pregnancy complications, and route of delivery (Table 3, Model 3). A similar but less elevated likelihood for LBW (AOR=1.45, 95% CI: 1.30–1.62), PTB (AOR=1.26, 95% CI: 1.13–1.40), and SGA (AOR=1.44, 95% CI: 1.30–1.59) was also observed among HIV-infected young mothers. Uninfected women of advanced maternal age did not achieve significant results for any of the observed feto-infant morbidities in the adjusted model.

In our study population, a significant incremental trend in both LBW and PTB was observed as maternal age increased, regardless of HIV status (p<0.01); however, probabilities for both of these adverse outcomes were higher among HIV-positive mothers than for those who were HIV-negative mothers (Fig. 1).

FIG. 1.

Trend in the probability of (a) low birth weight and (b) preterm births among HIV-infected and noninfected women by maternal age (Florida, 1998–2007).

Discussion

This study explored the impact of maternal HIV status and age on birth outcomes using a population-based dataset. Women who were of advanced maternal age and HIV positive had significantly increased odds of LBW, PTB, and SGA births, compared to young women without HIV infection. The association of these adverse birth outcomes (LBW,^8,9 PTB,^9,11,22 and SGA^8,22,23) with maternal HIV infection has been reported previously. The impact of maternal age on birth outcomes has been also previously studied.^2
–4,7,24 However, the combined effect of advanced maternal age and HIV infection on birth outcomes has been relatively unexplored. A prior study by Brown and colleagues found that HIV-positive women of advanced maternal age had a significantly higher likelihood of preterm delivery compared to HIV-negative women (p=0.0016).¹³ This study expands on the findings of Brown et al. by providing convincing evidence that maternal HIV infection is a risk factor not only for PTB, but also for LBW and SGA, with more pronounced risk for infants born to women of advanced age. The mechanisms explaining the association between maternal age and HIV infection with adverse birth outcomes are not clear, but may include demographic and lifestyle factors, as well as pregnancy-related clinical conditions. However, since our estimates were based on a matched sample, the association could not be explained by these factors alone. Moreover, the tests for adjusted trend remained significant, even in the matched sample.

After learning their HIV infection status, women and their families frequently have questions regarding the decision to have a child. This provides opportunities for crucial targeted interconception care for the reduction of feto-infant morbidities within this high-risk group.^25,26 Clinicians should consider the potential risk of adverse infant outcomes, such as LBW, PTB, and SGA, among women with HIV infection, particularly those who are within the advanced maternal age group. Present recommendations for interconception care include HIV screening of all women prior to pregnancy to reduce the risk of vertical disease transmission to infants and to provide counseling on health practices to reduce the likelihood of associated morbidities and mortalities.^26,27 Based on identified risk factors, such as HIV status and advanced maternal age, clinicians can provide tailored care and interventions to support healthy outcomes. Further research is warranted to strengthen the evidence base to support this assertion.

Our study does have some limitations. The utilization of an existing dataset results in the restriction of potential variables available for use and a possible lack of relevant information. For example, the dataset utilized in this study lacks information on the use of antiretroviral therapy or protease inhibitors among pregnant women. However, our findings are based on a propensity score weighted matching approach that estimates treatment effects by addressing possible bias in group assignment.²⁸ Our decision to dichotomize most variables included in the multivariable analysis has been criticized by some investigators due to potential power loss and bias in medical research analyses.^29,30 Furthermore, this decision may have led to the misclassification of some covariates. For instance, mothers with missing information for years of education completed were considered to have not completed high school; likewise, mothers with missing marital status information were categorized as unmarried. However, it is unlikely that this a priori decision impacted our study results, as a sensitivity analysis that excluded women with missing information for these variables yielded similar results (data not shown).

Despite these limitations, there are several strengths to this study. The large population size provides substantial power to detect differences between groups in the analysis. In addition, due to the population-based nature of the data, the likelihood of selection bias is minimal compared to single health facility-based data. Moreover, our reported estimates are based on a propensity score weighted matching approach, which minimizes selection bias. Hence, the findings of this study can be considered reliable and generalizable.

Given the potential heightened risk of poor birth outcomes among HIV-infected women of advanced maternal age, it is essential to design and implement relevant preventive and supportive strategies that could eliminate or reduce the burden. Policy and decision-makers need to be informed of this disparity in order to extend risk-reduction programs and services that address the unique needs of this at-risk group. Early detection of HIV infection is critical in an effort to prevent risky conception and adverse pregnancy outcomes, such as LBW, PTB, and SGA. Furthermore, due to the multifactorial nature of these outcomes, comprehensive public health strategies are required. This study provides further evidence of the importance of preconception and interconception care to improve pregnancy outcomes for women and their families through targeted intervention and counseling.

Footnotes

Author Disclosure Statement

No competing financial interests exist.

References

Cunningham

, Williams

. Williams Obstetrics [electronic resource] 23rd

Cunningham

et al. McGraw-Hill Medical: New York, 2010.

Balasch

, Gratacos

. Delayed childbearing: Effects on fertility and the outcome of pregnancy. Fetal Diagn Ther, 2011; 29:263–273.

Jolly

, Sebire

, Harris

, Robinson

, Regan

. The risks associated with pregnancy in women aged 35 years or older. Hum Reprod, 2000; 15:2433–2437.

Joseph

, Allen

, Dodds

, Turner

, Scott

, Liston

. The perinatal effects of delayed childbearing. Obstet Gynecol, 2005; 105:1410–1418.

Schempf

, Branum

, Lukacs

, Schoendorf

. Maternal age and parity-associated risks of preterm birth: Differences by race/ethnicity. Paediatr Perinat Epidemiol, 2007; 21:34–43.

Luke

, Brown

. Elevated risks of pregnancy complications and adverse outcomes with increasing maternal age. Hum Reprod, 2007; 22:1264–1272.

Odibo

, Nelson

, Stamilio

, Sehdev

, Macones

. Advanced maternal age is an independent risk factor for intrauterine growth restriction. Am J Perinatol, 2006; 23:325–328.

Bodkin

, Klopper

, Langley

. A comparison of HIV positive and negative pregnant women at a public sector hospital in South Africa. J Clin Nurs, 2006; 15:735–741.

Lambert

, Watts

, Mofenson

et al. Risk factors for preterm birth, low birth weight, and intrauterine growth retardation in infants born to HIV-infected pregnant women receiving zidovudine. Pediatric AIDS Clinical Trials Group 185 Team. AIDS, 2000; 14:1389–1399.

10.

Salihu

, Stanley

, August

, Weldeselasse

, Mbah

, Whiteman

. The association between HIV/AIDS during pregnancy and fetal growth parameters in Florida: A population based study. Curr HIV Res, 2012; 10:539–545.

11.

Marazzi

, Palombi

, Nielsen-Saines

et al. Extended antenatal use of triple antiretroviral therapy for prevention of mother-to-child transmission of HIV-1 correlates with favorable pregnancy outcomes. AIDS, 2011; 25:1611–1618.

12.

Iqbal

, Kriebs

, Harman

et al. Predictors of fetal growth in maternal HIV disease. Am J Perinatol, 2010; 27:517–523.

13.

Brown

, Holland

, Mosquera

, Calilap

, Bardeguez

. Human immunodeficiency virus infection in advanced maternal age gravidas. AIDS Res Hum Retroviruses, 2012; 28:265–269.

14.

McCowan

, Horgan

. Risk factors for small for gestational age infants. Best Pract Res Clin Obstet Gynaecol, 2009; 23:779–793.

15.

Taffel

, Johnson

, Heuser

. A method of imputing length of gestation on birth certificates. Vital Health Stat, 1982; 2:1–11.

16.

Alexander

, Kotelchuck

. Quantifying the adequacy of prenatal care: A comparison of indices. Public Health Rep, 1996; 111:408–418discussion 19.

17.

Kotelchuck

. An evaluation of the Kessner Adequacy of Prenatal Care Index and a proposed Adequacy of Prenatal Care Utilization Index. Am J Public Health, 1994; 84:1414–1420.

18.

Alexander

, Allen

. Conceptualization, measurement, and use of gestational age. I. Clinical and public health practice. J Perinatol, 1996; 16:53–59.

19.

Dehejia

, Wahba

. Propensity score-matching methods for nonexperimental causal studies. Rev Econ Stat, 2002; 84:151–161.

20.

Austin

. A critical appraisal of propensity-score matching in the medical literature between 1996 and 2003. Stat Med, 2008; 27:2037–2049.

21.

Breslow

, Day

. Statistical Methods in Cancer Research: The Analysis of Case-Control Studies. International Agency for Research on Cancer: Lyon, France, 1980.

22.

Habib

, Daltveit

, Bergsjo

, Shao

, Oneko

, Lie

. Maternal HIV status and pregnancy outcomes in northeastern Tanzania: A registry-based study. BJOG, 2008; 115:616–624.

23.

Ndirangu

, Newell

M-L

, Bland

, Thorne

. Maternal HIV infection associated with small-for-gestational age infants but not preterm births: Evidence from rural South Africa. Hum Reprod, 2012; 27:1846–1856.

24.

Bakker

, Steegers

EAP

, Biharie

, Mackenbach

, Hofman

, Jaddoe

VWV

. Explaining differences in birth outcomes in relation to maternal age: The Generation R Study. BJOG, 2011; 118:500–509.

25.

Moore

, Parrish

, Black

. Interconception care for couples after perinatal loss: A comprehensive review of the literature. J Perinat Neonatal Nurs, 2011; 25:44–51.

26.

Johnson

, Posner

, Biermann

et al. Recommendations to improve preconception health and health care–United States. A report of the CDC/ATSDR Preconception Care Work Group and the Select Panel on Preconception Care. MMWR Recomm Rep, 2006; 55:1–23.

27.

Jack

, Atrash

, Coonrod

, Moos

, O'Donnell

, Johnson

. The clinical content of preconception care: An overview and preparation of this supplement. Am J Obstet Gynecol, 2008; 199:S266–279.

28.

Diaz

, Handa

. An assessment of propensity score matching as a nonexperimental impact estimator. J Hum Resources, 2006; XLI:319–345.

29.

Naggara

, Raymond

, Guilbert

, Roy

, Weill

, Altman

. Analysis by categorizing or dichotomizing continuous variables is inadvisable: An example from the natural history of unruptured aneurysms. AJNR Am J Neuroradiol, 2011; 32:437–440.

30.

Royston

, Altman

, Sauerbrei

. Dichotomizing continuous predictors in multiple regression: A bad idea. Stat Med, 2006; 25:127–141.