Women’s Health After Pregnancy and Child Outcomes at Age 3 Years: A Prospective Cohort Study

Abstract

Objectives. This study examined the persistence and comorbidity of women’s physical and mental health conditions after pregnancy and the association of these conditions with child outcomes.

Methods. A national cohort of women who recently gave birth were surveyed in 1988 and again in 1991. We examined longitudinal data on maternal poor physical health, depressive symptoms, and smoking, and maternal report of child outcomes (at age ∼3 years).

Results. Women’s poor physical health and smoking had strong, graded associations with children’s physical health and behavior problems, whereas women’s depressive symptoms were associated with children’s delayed language and behavior problems.

Conclusions. Substantial persistence and comorbidity of women’s health conditions exist after pregnancy with adverse effects on early child outcomes. Child health professionals should support services and policies that promote women’s health outside the context of pregnancy.

Research on the relation between the health of women and that of their children has traditionally focused on health conditions that arise in pregnancy and the early perinatal period. This emphasis has profoundly shaped the nature of maternal and child health services and policy in the United States.1–3 Emerging research, however, highlights the additional role of women’s health both before and after pregnancy as a determinant of child health and well-being. For example, the presence of chronic illnesses, poor nutritional status, and health risk behaviors in women before pregnancy appears to strongly influence neonatal morbidity and mortality.4,5 This research has led to an increasing focus on health services for women before pregnancy.6–8

Significant gaps exist in research linking women’s health after pregnancy and early child outcomes. First, most studies have sought to isolate the effects of single maternal risk factors, such as smoking.9,10 There has been little effort to assess more comprehensively the interrelated conditions women experience after delivery and the scope of their influence on child outcomes. For example, despite the well-documented association between adult smoking and depression,11 little is known about their joint effect on child health. Furthermore, most studies have relied on cross-sectional assessments of the maternal and child health relationship. More longitudinal approaches are needed to clarify the continuity or persistence of conditions over time and to provide accurate estimates of effects.12 For example, longitudinal models have better delineated the effect of prolonged exposure to poverty on both birth outcomes and child development.13,14 The disadvantage of a focus on isolated maternal risk factors, assessed at a single point in time, is evident in the multiple distinct and unrelated calls for action on the part of child health professionals.15–20 Little coherent support among pediatricians has been generated for more comprehensive and continuous health care for women after they deliver.21–24

The goal of this study was to offer a broader perspective on the scope and persistence of the maternal and child health relationship after pregnancy as an empirical first step in reconsidering current health programs and policies for women and their children. Specifically, we used a longitudinal cohort to examine maternal poor physical health, depressive symptoms, and smoking at 2 points in time after delivery and their association with child health and development at age 3 years. We also examined social and economic factors that may frame the distribution of maternal conditions.

METHODS

Data were obtained from the 1988 National Maternal and Infant Health Survey and its 1991 Longitudinal Follow-Up.25,26 The 1988 National Maternal and Infant Health Survey enrolled a national probability sample of 9953 births and collected data on maternal and infant health, health service use, and demographic information approximately 17 ± 5 months after delivery. Black infants and low-birthweight infants were oversampled. The 1991 Longitudinal Follow-Up to the National Maternal and Infant Health Survey successfully recontacted 8285 (83%) families by telephone or personal interview when the child was aged 35 ± 5 months. A total of 8087 children still living with their mothers at the time of the second interview formed the basis for our study.

Child Outcomes

Child outcomes were from the 1991 Longitudinal Follow-Up survey and reflect physical health, language, and behavior outcomes. Child physical health outcomes, as reported by the mother, included the child’s general health status (dichotomized as excellent/very good/good vs fair/poor), having ever been hospitalized (not including at birth), and the mother having ever been told by a health care provider that the child had asthma. Language outcomes included maternal report (yes/no) as to whether the child had ever counted 3 objects correctly, said the names of at least 4 colors, and counted out loud to 10. Child behavior was assessed with 3 questions: whether the child was difficult to manage most of the time (vs easy to manage/sometimes difficult), whether the child had some or great difficulty playing with other children (vs got along well all or most of the time), and whether the child had frequent tantrums (vs never/occasional tantrums).

Women’s Health

Women’s health was characterized by self-reported physical health status, depressive symptoms, and smoking. Based on data from both the 1988 and the 1991 surveys, the health states were categorized as present both times, present 1 time, or absent.27 From prior work on women’s physical health after delivery,28 physical health was characterized as poor in 1988 if a mother reported being rehospitalized or having had 3 or more visits to a doctor in the interval since delivery (17 ± 5 months) and as poor in 1991 if a mother reported her general health to be fair or poor (vs excellent/very good/good) at the time of the second survey. Depressive symptoms were ascertained with the 20question Center for Epidemiological Studies Depression Scale administered in both the 1988 and the 1991 surveys. A score of 16 or greater is considered positive in screening for depression.29 Maternal smoking was assessed with a single question (yes/no) both in 1988 and in 1991. Prior studies have shown the correlation between self-reported maternal smoking and infant urine cotinine level to be good.30

Sociodemographic and Other Covariates

In the 1988 survey, mothers were asked to report on their total family income in the 12 months before delivery. To adjust for family size, a ratio of income to poverty line income to poverty line was created based on the 1988 federal poverty thresholds. For example, an income of $6000 for a family of 4 was divided by the federal threshold of $12 092 to yield a ratio of 0.5. The ratio was then categorized into quintiles (< 0.9, 0.9–1.9, 2.0–3.3, 3.4–4.8, > 4.8). Other covariates included the mother’s age (< 20 years, 20–29 years, 30 years and older), highest attained education (< high school, high school or general equivalency diploma, ≥ some college), marital status (married, separated/divorced, single), race/ethnicity (White non-Hispanic, Black non-Hispanic, Hispanic), birthweight (< 1500 g, 1500–2499 g, ≥ 2500 g), and the child’s age at the time of the 1991 survey. Additional variables used to adjust for the infant’s health status at birth included the presence of a major birth defect (spina bifida, heart condition, cerebral palsy, hydrocephalus, sickle cell anemia, or other genetic disorder), history of hospital transfer at birth, and rehospitalization in the first month after discharge.

Analysis

Bivariate relationships between maternal health and child outcomes were assessed with χ² analyses. Logistic regression was used for multivariate analyses. Variables for the 3 maternal health conditions were entered together into the models. This ensured, for example, that the association of child outcomes with maternal physical health was estimated independently of any comorbid depressive symptoms. Potential interactions between comorbid maternal conditions were explored. All multivariate models were adjusted for maternal age, income, education, marital status, and race/ethnicity and for child birthweight and age. Finally, to address the potential for reverse causation (i.e., that children’s poor health at birth might be the primary determinant of maternal conditions in the next 3 years), a final set of regressions reexamined these relationships in a restricted sample of infants most likely to be healthy at birth (≥ 2500 g, no history of major birth conditions, no hospital transfer after delivery, and no rehospitalization in the first month after discharge). All analyses were weighted to account for the complex survey design and for nonresponse; weights were provided by the National Center for Health Statistics. Analyses were conducted with SUDAAN, Version 7.5.3 (Research Triangle Institute, Research Triangle Park, NC).

RESULTS

Sociodemographic and Health Characteristics

The sociodemographic characteristics of the cohort reflect a nationally representative sample of women giving birth in 1988.26 Seventy-one percent of the women were nonHispanic White, 17% were non-Hispanic Black, and 12% were Hispanic; by the time of the 1991 survey, 46% had at least some college education, and 37% had a high school degree. At the time of child assessment in the 1991 survey, 82% of the children were aged 2.5 to 3.5 years.

The prevalence of each health condition is shown in Table 1 ▶. Among the women, depressive symptoms were the most commonly reported condition and smoking was the most likely to be reported both times. Approximately 60% of the women reported at least 1 of these conditions. Comorbidity was common, with 24% of the women ever reporting 2 or more conditions. For example, more than 40% of the women with persistent smoking and more than 50% of the women with persistent poor physical health also reported depressive symptoms in at least 1 time period. Twenty-two percent of the mothers reported that their child had 1 or more poor physical health outcomes, 51% reported that their child had not reached 1 or more of the language milestones, and 25% noted that their child had 1 or more of the behavior problems.

TABLE 1.

—Health of Women and Children in 1988 National Maternal and Infant Health Survey and Its 1991 Longitudinal Follow-Up

	N	Weighted %
Women’s health during years after delivery	8087
Poor physical health
Both times	257	2.3
1 time	2227	23.6
Absent	5603	74.1
Depressive symptoms
Both times	981	9.0
1 time	2339	25.4
Absent	4767	65.6
Smoking
Both times	1855	22.1
1 time	691	7.8
Absent	5541	70.1
Child health and development at age 3 years
Physical health
Fair or poor health	392	3.7
Ever been hospitalized	1638	16.2
Ever been diagnosed with asthma	840	7.2
Language
Cannot count to 10	3648	43.7
Cannot count 3 objects	1293	14.3
Cannot name 4 colors	2342	23.0
Behavior
Difficult to manage most of the time	450	4.7
Frequent tantrums (shouting, screaming, kicking)	1178	12.3
Some or great difficulty in getting along with others	1230	14.6

Antecedent Income and Women’s Health

A strong gradient existed between income quintile in the 12 months before delivery and persistence of maternal poor physical health, depressive symptoms, and smoking in the 3 years after delivery (Figure 1 ▶). Compared with women in the top income quintile, women in the bottom quintile were much more likely to report poor physical health (4.6% vs 0.4%, P < .001), depressive symptoms (16.9% vs 2.8%, P < .001), and smoking (31.5% vs 11.3%, P < .001) in both the 1988 and the 1991 surveys. The increased prevalence of comorbidity also was associated lower with income.

FIGURE 1.

—Women’s persistent poor physical health, depressive symptoms, and smoking, by income quintile: 1988 National Maternal and Infant Health Survey and 1991 Longitudinal Follow-Up.

Women’s Health After Delivery and Child Outcomes at Age 3 Years

In bivariate analyses, women’s poor physical health, depressive symptoms, and smoking were each associated with multiple maternal-reported child adverse outcomes. The persistence of maternal conditions provided evidence for a graded association with child outcomes (Table 2 ▶). Women’s poor physical health and smoking status were more strongly associated with children’s poor physical health and behavior problems than with language problems, whereas women’s depressive symptoms were associated with all child physical health, language, and behavior problems. A dose–response gradient was evident, for example, among children whose mothers reported depressive symptoms. Children of women reporting depressive symptoms both times vs 1 time vs never had significantly different poor outcomes, including rates of fair or poor child health (9.7% vs 4.9% vs 2.5%, P < .001), not being able to name 4 colors (35.3% vs 27.7% vs 19.5%, P < .01), and having frequent tantrums (25.5% vs 16.9% vs 8.7%, P < .001).

TABLE 2.

—Child Health and Development at Age 3 and Women’s Physical Health, Depressive Symptoms, and Smoking During First 3 Years After Delivery: Bivariate Results

	Child Physical Health			Child Language			Child Behavior
Women’s Health	Fair/Poor Health	Ever Hospitalized	Reported Asthma	Not Counting to 10	Not Counting 3 Objects	Not Naming 4 Colors	Difficult Most of Time	Frequent Tantrums	Some/Great Difficulty Getting Along
Poor physical health
Both times	22.3	29.8	14.5	59.0	21.4	36.4	14.2	25.0	19.3
1 time	6.3	18.2	8.4	45.0	13.1	24.0	6.9	14.6	17.5
Absent	2.4	15.2	6.6	42.8	14.5	22.3	3.7	11.1	13.5
P	.001	.001	.01	.01	NS	.01	.001	.001	.01
Depressive symptoms
Both times	9.7	22.4	11.0	56.2	20.8	35.3	11.5	25.5	24.9
1 time	4.9	17.3	8.1	48.9	17.3	27.7	7.2	16.9	18.8
Absent	2.5	15.0	6.3	40.0	12.3	19.5	2.8	8.7	11.8
P	.001	.001	.01	.001	.001	.001	.001	.001	.001
Smoking
Both times	4.7	21.5	10.6	48.4	15.4	22.9	6.1	17.9	18.0
1 time	5.5	19.6	8.0	45.5	14.1	23.9	6.6	16.5	16.0
Absent	3.2	14.2	6.0	42.0	14.0	22.9	4.0	10.0	13.3
P	.05	.001	.001	.01	NS	NS	.05	.001	.01

Note. NS = not significant.

Source. Data are from the 1988 National Maternal and Infant Health Survey and the 1991 Longitudinal Follow-Up.

Logistic regression models examined these same relationships, adjusting for known covariates, including income, maternal education, race/ethnicity, age, marital status, and child birthweight and age (Table 3 ▶). Only the relation between maternal depressive symptoms and child physical health was substantially attenuated. Evidence for the graded associations remained, with persistent maternal conditions having a greater negative effect than intermittent ones.

TABLE 3.

—Child Health and Development at Age 3 and Women’s Physical Health, Depressive Symptoms, and Smoking During First 3 Years After Delivery: Logistic Regression^a

	Child Physical Health			Child Language			Child Behavior
Women’s Health	Fair/Poor Health Adjusted OR (95% CI)	Ever Hospitalized Adjusted OR (95% CI)	Reported Asthma Adjusted OR (95% CI)	Not Counting to 10 Adjusted OR (95% CI)	Not Counting 3 Objects Adjusted OR (95% CI)	Not Naming 4 Colors Adjusted OR (95% CI)	Difficult Most of Time Adjusted OR (95% CI)	Frequent Tantrums Adjusted OR (95% CI)	Some/Great Difficulty Getting Along Adjusted OR (95% CI)
Poor physical health
Both times	7.1 (4.1, 12.2)	1.8 (1.2, 2.9)	1.8 (1.0, 3.0)	1.3 (0.9, 2.0)	1.2 (0.7, 2.1)	1.3 (0.8, 2.1)	2.4 (1.3, 4.1)	1.6 (1.0, 2.4)	1.1 (0.7, 1.7)
1 time	2.3 (1.7, 3.2)	1.2 (1.0, 1.4)	1.2 (0.9, 1.5)	1.0 (0.9, 1.2)	0.8 (0.7, 1.0)	0.9 (0.8, 1.1)	1.5 (1.1, 2.0)	1.1 (0.9, 1.3)	1.2 (1.0, 1.4)
Absent	1.0	1.0	1.0	1.0	1.0	1.0	1.0	1.0	1.0
Depressive symptoms
Both times	2.2 (1.4, 3.4)	1.2 (0.9, 1.6)	1.2 (0.9, 1.7)	1.5 (1.2, 1.9)	1.8 (1.4, 2.4)	1.8 (1.4, 2.2)	3.0 (2.0, 4.5)	2.6 (2.0, 3.4)	2.1 (1.6, 2.7)
1 time	1.3 (0.9, 1.9)	1.0 (0.8, 1.2)	1.0 (0.8, 1.4)	1.3 (1.1, 1.5)	1.5 (1.2, 1.8)	1.4 (1.2, 1.6)	2.1 (1.5, 3.0)	1.7 (1.4, 2.1)	1.5 (1.2, 1.9)
Absent	1.0	1.0	1.0	1.0	1.0	1.0	1.0	1.0	1.0
Smoking
Both times	1.0 (0.7, 1.4)	1.4 (1.2, 1.7)	1.6 (1.2, 2.1)	1.0 (0.8, 1.2)	0.9 (0.8, 1.2)	0.8 (0.7, 1.0)	1.2 (0.9, 1.7)	1.7 (1.4, 2.1)	1.3 (1.0, 1.6)
1 time	1.4 (0.9, 2.5)	1.4 (1.0, 1.8)	1.3 (0.9, 1.8)	1.0 (0.8, 1.3)	1.0 (0.7, 1.4)	1.0 (0.7, 1.3)	1.4 (0.9, 2.3)	1.6 (1.2, 2.1)	1.1 (0.8, 1.5)
Absent	1.0	1.0	1.0	1.0	1.0	1.0	1.0	1.0	1.0

Note. OR = odds ratio; CI = confidence interval. Values in boldface type are significant at P < .05.

Source. Data are from the 1988 National Maternal and Infant Health Survey and the 1991 Longitudinal Follow-Up.

^aAll models control for income, maternal education, race/ethnicity, age, marital status, and child birthweight and age.

Interaction terms were examined to assess the combined effects of comorbid maternal conditions on child outcomes, but these effects were nonsignificant. However, we tested the assumption that the independent variables (i.e., the maternal conditions) in a logistic regression model have a multiplicative relationship with one another.31 For example, based on Table 3 ▶, the odds of a child’s being “difficult to manage” if the mother has both intermittent poor physical health (odds ratio [OR] = 1.5) and persistent depressive symptoms (OR = 3.0) should reflect the product of the 2 odds (i.e., OR = 4.5). We examined this by substituting these 2 independent maternal conditions with a single variable to reflect comorbidity, and the odds ratio was approximately the same (4.2; 95% confidence interval [CI] = 2.1, 8.2). Furthermore, among women with both persistent poor physical health and persistent depressive symptoms, the odds of a child’s being difficult to manage and having frequent tantrums were 9.0 (95% CI = 4.1, 20.2) and 4.2 (95% CI = 2.2, 8.3), respectively, compared with children of women who never reported poor physical health or depressive symptoms.

To address the potential for reverse causation (i.e., that poor child health at birth primarily determined maternal health in the subsequent 3 years), the analyses were rerun but restricted to infants most likely to be healthy at birth (i.e., normal birthweight, no major birth conditions, no hospital transfer at birth, and no rehospitalization in the first month after discharge). There was no change in the associations between maternal health conditions and child health outcomes.

DISCUSSION

These findings show the persistence and comorbidity of women’s health conditions well after pregnancy and their adverse associations with multiple child outcomes at age 3 years. Moreover, the results suggest that the more persistent the woman’s poor physical health, depressive symptoms, and smoking, the greater the negative effects on child health. Effects previously reported in cross-sectional studies of isolated risk factors appear in fact to operate together, over time, and in a potentially multiplicative manner. These results suggest the need to develop new intervention strategies to optimize health outcomes for women and children.

Our findings with a national cohort confirm and extend previous studies drawn from a variety of clinical populations.28,32,33 Crosssectional surveys of women bringing their children for pediatric care have found that 15% to 40% report depressive symptoms and that 9% to 40% smoke, depending on the demographic characteristics of the clinic sample.32,33 The substantial comorbidity seen in this study is similar to that found in a prevalence study of 8 health conditions in women with children younger than 18 months.33 Much less work has been done on the persistence of these maternal conditions. Haas and McCormick28 investigated poor physical health among women who had delivered low-birthweight infants. They found that 16.9% of the women reported fair or poor health and that almost 30% were hospitalized in the 5 years after delivery. The persistence of depressive symptoms found in our nationally representative sample is similar to that of another recent large study of mothers with young children.34

The strong gradient between income before delivery and comorbidity and persistence of negative maternal health conditions in the years following delivery is striking, especially in the context of the prevalence of low income among childbearing women.35 Acknowledging the social and structural forces that shape the distribution of these conditions is critical for understanding the upstream causes of health disparities and for developing effective interventions.36 Interventions for maternal smoking that implicitly emphasize the role of individual volition may fail owing to lack of recognition of the forces structuring women’s options (e.g., access to nicotine patches).37 Similar failures may result from inability to appreciate the structural barriers faced by women with depression (e.g., access to affordable mental health services).

Although the specific pattern of maternal and child health associations we found is supported by a constellation of prior studies, the scope and persistence of these relationships have not been previously reported. Maternal smoking has been linked to childhood asthma, hospitalization, and behavior problems,9,10 and maternal depression is known to be associated with child language development34 and behavior problems,20 thus offering some support for the validity of our conceptual framework and outcome measures. That the relationships span multiple maternal conditions and cut across various domains of child health and development suggests that a common interest in ensuring women’s comprehensive health may emerge from diverse clinical disciplines. Our findings also imply that interest in more narrowly framed conditions, such as postpartum depression, should take a more longitudinal perspective and should assess the additional burden imposed by other co-occurring maternal morbidities. Taken together, this broad intergenerational effect on health in the first years after delivery may help to inform recent findings on the capability of early parental socioeconomic status and well-being to alter children’s health trajectories over the life course.37,38

With respect to interventions, our findings suggest that highly targeted initiatives to address maternal health problems may not be adequate to alter outcomes for many women and children. Findings on maternal depression, smoking, folic acid levels, and domestic violence, among others, have led to a series of calls for pediatric action. Given the persistence and comorbidity of these problems, however, a more integrated strategy to link women to adult clinicians or social agencies may be needed.

Such a strategy might involve use of child-oriented settings as gateways to care for women. These settings might include, for example, pediatric clinics and Special Supplemental Nutrition Program for Women, Infants, and Children clinics, as well as Early Intervention and Head Start programs. In the health care setting, this approach would complement the work of family practice in ensuring comprehensive health care for families.39 These gateways or links could be built on the strengths of specific sites, such as the collocation of adult and pediatric clinicians or the integrated information systems of health maintenance organizations. Recent state efforts to expand Medicaid and State Children’s Health Insurance Program eligibility for parents may reflect an important step toward reducing financial barriers to such a strategy.3,40,41 Continuous health care also should address women’s health well before any pregnancy; substantial work remains to be done in this regard.

Several considerations should frame the interpretation of these findings. Health information on the cohort was limited to 2 time points separated by approximately 18 months. Conditions present at both times actually may have been recurrent rather than persistent. The persistence or recurrence of these conditions and their effect on child health need to be more fully delineated. However, few longitudinal data sets include multiple measures of women’s health and child outcomes.42 Similarly, we know of no longitudinal data sets with which to address the role of paternal health. Women responded to questions on their own behalf and on behalf of their children, raising the possibility that women in poor physical or mental health may have simply perceived their children to be in worse health. That women with persistent depressive symptoms were not more likely to report their children to have poor physical health, however, makes this possibility somewhat less likely. Future research should include the use of nonparental informants and validated assessment tools.

The temporal sequence in which the variables were measured is a relative strength of the analysis, but reverse causation cannot be ruled out. Poor health status at birth may have led to both maternal depression and adverse child outcomes at age 3 years. However, when analyses were restricted to normalbirthweight infants without major birth conditions or early rehospitalization, the findings were unchanged. Nevertheless, complex relationships likely operate in which the poor health of the mother and the child become reinforcing. Taken together, the consistency, strength, and graded nature of the associations found, along with their temporal sequence and biological plausibility, make it unlikely that these limitations substantially undermine the results.

This study focused on the relevance of women’s health to children’s outcomes. Clearly, these same health conditions adversely affect women’s capacity in roles other than parenting. The degree to which adults’ poor physical and mental health undermine successful employment and social relationships, as well as long-term health trajectories, suggests that demands for improved health services for women need not be premised solely on their benefits for child outcomes. Rather, these findings should serve to encourage the child health community to join women’s health advocates in promoting women’s access to comprehensive and continuous health services.

Our data suggest that inequalities in the health of women may be a primary contributor to health disparities in young children and that these intergenerational effects on health are not confined to pregnancy. Although the child health community has long supported the provision of health services to women while they are pregnant, these findings imply the need to expand that commitment beyond the prenatal period. Such a commitment might begin with efforts to configure practical health service, health insurance, and public health initiatives that ensure optimal outcomes for both women and their children.