| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Touch Research Institutes, Department of Pediatrics, University of Miami School of Medicine, Miami, Florida (M.A.D., T.F., M.H.-R.); the Department of Psychology, Florida Atlantic University (N.A.J.); Fielding Graduate University (T.F.); the Department of Pharmacology, Duke University Medical Center (S.S., C.K.); and the Department of Obstetrics and Gynecology, University of Miami School of Medicine, Miami, Florida (A.G.-G.).
Address correspondence and reprint requests to Miguel A. Diego, PhD, Touch Research Institutes, Department of Pediatrics, University of Miami School of Medicine, P.O. Box 016820. Miami, FL 33101. E-mail: mdiego{at}med.miami.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Methods: Fetal ultrasound biometry measurements and maternal sociodemographic characteristics, emotional distress symptoms, and first morning urine samples were collected during a clinical ultrasound examination for a cross-sectional sample of 98 women who were between 16 and 29 weeks pregnant. Fetal weight was estimated from ultrasound biometry measurements; maternal emotional distress was assessed using the daily hassles (stress), Center for Epidemiologic Studies–Depression (depression), and State-Trait Anxiety Inventory (anxiety) scales; and urine samples were assayed for cortisol and norepinephrine levels.
Results: Correlation analyses revealed that both maternal psychological (daily hassles, depression, and anxiety) and biochemical (cortisol and norepinephrine) variables were negatively related to fetal biometry measurements and estimated fetal weight. A structural equation model further revealed that when the independent variance of maternal sociodemographic, psychological distress, and biochemistry measures were accounted for, prenatal cortisol was the only significant predictor of fetal weight.
Conclusions: Women exhibiting psychological distress during pregnancy exhibit elevated cortisol levels during midgestation that are in turn related to lower fetal weight.
Key Words: depression • anxiety • stress • cortisol • norepinephrine • fetal weight
Abbreviations: CRH = corticotropin-releasing hormone; HPA = hypothalamic–pituitary axis; IGFBP1 = insulin-like growth factor binding protein 1; eFW = estimated fetal weight; BPD = biparietal diameter; AC = abdominal circumference; FL = femur length; HC = head circumference; SES = socioeconomic status; SEM = structural equation model; CFI = comparative fit index; RMSEA = root mean square error of approximation.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Prenatal maternal distress effects on the fetus appear to be mediated by maternal neuroendocrine function. Psychological distress during pregnancy has been associated with both the hypothalamic–pituitary axis (HPA) (i.e., elevated cortisol and corticotropin-releasing hormone [CRH]) and sympathoadrenal dysregulation (i.e., elevated norepinephrine levels and cardiovascular function) (4,6,9,10,14–16). Data from a wide range of studies indicate that maternal HPA axis and sympathoadrenal function during pregnancy affect fetal development. For example, maternal HPA dysregulation (i.e., elevated cortisol/CRH levels) has been associated with premature delivery and low birthweight (4,6,9,10,16,17). Research suggests that corticosteroid infusions to the mother result in up to a 25% reduction in fetal weight (18). Similarly, maternal sympathoadrenal dysregulation (i.e., elevated maternal norepinephrine levels) has been associated with low birthweight (4), and norepinephrine infusions induce uterine artery contractions and decrease uterine artery blood flow in humans (19,20). Furthermore, norepinephrine administered to pregnant ewes affects uterine artery and placental blood flow and decreases fetal insulin concentrations, increases fetal insulin-like growth factor binding protein 1 (IGFBP-1) concentrations and results in fetal growth restriction (21,22). Although studies have been conducted to assess the effects of maternal HPA and sympathoadrenal function on fetal growth in animal models and birthweight and prematurity in humans, we are not aware of any studies that have assessed the effects of prenatal cortisol and norepinephrine on human fetal weight.
The present study examined the effects of maternal psychological distress during midgestation on fetal development by assessing the relations between fetal growth measures and maternal stress, depression, and anxiety during midgestation. Midgestation fetal measurements rather than neonatal measurements were collected to provide a more proximate assessment of the effects of prenatal maternal distress during midgestation. The study also tested for mediating effects on the relations between maternal psychological distress, cortisol, and norepinephrine on fetal weight. Based on previous research indicating that stressed, depressed, and anxious mothers are at greater risk for premature delivery and low birthweight, maternal psychological distress variables were expected to be negatively related to fetal weight. Inasmuch as the effects of maternal psychological distress on prematurity and low birthweight appear to be mediated by maternal neuroendocrine function, prenatal cortisol and norepinephrine were also expected to be related to estimated fetal weight.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
2 tests failed to reveal any significant gestational age period (<19 weeks, 19–21 weeks, 22–24 weeks, >24 weeks) differences on maternal age, SES, ethnicity, marital status, or infant gender. To avoid potential errors and confounds associated with estimating age based on ultrasound measurements, gestational age was estimated based on the mothers’ last menstrual period (LMP). Only women who were able to accurately recall their LMP date were included in this study. Data were only collected for low-risk pregnancy mothers who: a) did not exhibit any pregnancy complication, including hypertensive disorders (preeclampsia), gestational diabetes, ectopic pregnancies, anemia, placenta previa, placental abruption, or intrauterine growth restriction; b) were not diagnosed with HIV or any other infectious disease; and c) did not report having any metabolic or eating disorder (obesity, bulimia, anorexia) or any psychiatric condition other than depression or any anxiety disorder. Women who reported smoking cigarettes, drinking alcohol, or using recreational drugs during pregnancy or taking medications (including antidepressant or other psychotropic medications) other than vitamins during pregnancy were excluded from the study. Approximately 60% of women met inclusion criteria. Of these, approximately 70% agreed to participate.
Assessments
Maternal questionnaires were administered during the same visit that the urine samples were obtained and clinical ultrasound examinations performed. A demographic questionnaire was administered to the mothers after they signed an informed consent. The questionnaire included the mothers’ marital status, age, ethnicity, occupation, and education. Ethnicity was dummy coded into 0 = white or 1 = not white, and renamed minority status and marital status was dummy-coded into 0 = single or 1 = having a significant other (husband or boyfriend). Answers to the occupation and education questions were used to compute SES based on the Hollingshead Four Factor Index of Social Status (23).
The Center for Epidemiological Studies–Depression scale (CES-D) (24) is a 20-item scale with scores ranging between 0 and 60. The respondents rate the frequency (within the last week) of 20 symptoms. The symptoms include depressed mood, feelings of helplessness and hopelessness, feelings of guilt and worthlessness, loss of energy, and problems with sleep and appetite. Although the CES-D does not allow for a clinical diagnosis of depression, it has adequate discriminant and convergent validity, correlates well with other depression scales, and can differentiate clinically depressed and nondepressed subjects with only a 6% false-positive and 36% false-negative rate (25). In addition, this scale has been shown to be reliable and valid for diverse demographic groups (25). Internal consistency for the CES-D is high (Cronbach’s alpha = 0.82–0.85) and test–retest reliability is satisfactory (r = 0.51–0.67 over a 2- to 8-week period; r = 0.32–0.54 over a 3- to 12-month period) (25). Sixty percent of the women met CES-D criteria for depression (a cutoff score of 16); they had a mean CES-D score of 22.4 (SD = 10.6; range, 1–46).
The Trait Anxiety Inventory (26) was used to assess anxiety. The Trait Anxiety Inventory consists of 20 items (each item scored on a 4-point Likert scale) that yields a summed score from 20 to 80. Higher scores indicate greater anxiety. The Trait Anxiety Scale is comprised of items on how the subject "typically feels," including "I feel nervous" and "I feel calm." The scores increase in response to stress and decrease under relaxing conditions (26). Research has demonstrated that the State-Trait Anxiety Inventory has adequate concurrent validity and internal consistency (r = 0.83) (26). The mothers had a mean Trait Anxiety Scale score of 42.4 (SD = 9.3; range, 23–68).
The Daily Hassles Scale (Field, unpublished scale) was used to assess daily hassles. This 14-question, 4-point Likert scale forms a summed score from 14 to 56. Higher scores indicate greater daily hassles with a score of 1 indicating "no hassle" and a score of 4 indicating a "big hassle." The Hassles Scale is comprised of items that can be perceived as routine hassles for the individual including people (i.e., friends, mother, landlord) and resources (i.e., getting money, food, time). The Daily Hassles Scale was used in the present study because this is a brief scale developed to assess stressors specific to pregnancy in middle to lower SES women from diverse ethnic and cultural backgrounds. Reliability computed on the hassles scale scores collected in this study revealed adequate internal consistency and reliability (Cronbach’s alpha = 0.91). The mothers had a mean Daily Hassles score of 22.7 (SD = 5.9; range, 12–39).
Urine Assays
Urine samples were collected from mothers on the day of the clinical ultrasound examination. No systematic differences were noted in the sampling or the timing of collection of urine samples between women. Because preservation of urine catecholamines with acid can lead to erroneously high catecholamine values (27), urine samples were transferred to plastic vials and frozen without using acid or other preservatives.
Urinary cortisol was assayed in the stored urine samples by radioimmunoassay using a specific antiserum from Radioassay Systems Laboratories (Carson City, CA). The specificity of the assay is such that biologic fluids can be assayed directly after heat inactivation of CBG, eliminating the need for time-consuming extraction into organic solvents, which is usually required for this assay. Only 5 to 10 µL of sample is needed for triplicate assay. Specially purified 3H-cortisol from the same supplier is used as the labeled hormone. Bound and free hormones are separated by the dextran-coated charcoal technique. The sensitivity of the assay is 0.025 ng per tube. The interassay and intraassay coefficient of variation is less than 10% and 5%, respectively. Standards are prepared from cortisol from the same supplier, and quality control samples representing low, medium, and high values are run in every assay. Cortisol values were corrected for creatinine volume.
Urinary norepinephrine was assayed by high-pressure liquid chromatography (HPLC) and electrochemical detection after online purification by cation exchange chromatography. Urine is injected directly onto the chromatograph, which contains an online cation exchange column. Samples are absorbed onto this precolumn using a mobile phase of low ionic strength and norepinephrine and normetanephrine arc eluted with a mobile phase of higher ionic strength. Norepinephrine is detected with a TL5A glassy carbon working electrode maintained at +600 mV. An internal standard is added at the beginning of the procedure. Standard curves are generated from catecholamine-free urine to which known amounts of standard have been added. Sensitivity of the assay is 2550 pg. Norepinephrine values were corrected for creatinine volume.
Fetal measures were obtained from the mothers’ clinical fetal ultrasound biometry records. Fetal ultrasound biometry measurements were performed by clinical ultrasonographers on an Aloka 5500 ultrasound machine with a 5-MHz curvilinear abdominal probe. Fetal femur length, head circumference, abdominal circumference, and biparietal diameter measurements were conducted using standard clinical measurement protocols (28–30).
The principal eight fetal weight estimation algorithms were then used to estimate fetal weight from fetal ultrasound biometry measurements (31). Fetal weight estimation algorithms based on fetal ultrasound biometry measurements provide sufficient parametric information to allow for the accurate reconstruction of the fetal volume. These algorithms produce comparably accurate estimates of fetal weight. Estimated fetal weight algorithms produce estimates with a 10.0% mean absolute percentage error. Fetal weight estimation is not significantly improved by the use of additional nonstandard sonographic fetal measurements or by the use of multiple assessments. There is no significant bias in fetal weight estimation introduced by differences in operator training or diagnostic setting (32). Furthermore, fetal growth is more accurately estimated through the use of estimated fetal weight equations than with individual ultrasound biometry measurements (33).
Fetal ultrasound biometry measures were highly correlated with fetal weight estimates (r = 0.89–0.97), suggesting that all of these measures describe a similar construct. Fetal weight estimated using eight different algorithms produced comparable values that showed a high degree of correlation (r = 0.91–0.99). Of these methods of fetal weight estimation, the Shephard formula (34) showed the highest correlation with the other estimated fetal weight formulas (r = 0.98–0.99). As such, values obtained using the Shephard formula were used as the estimate of fetal weight. The Shephard formula is widely used in clinical practice, because most commercial ultrasound machines readily provide it. This formula results in lower estimation errors than other formulas in fetuses less than 30 weeks’ gestation (35) or fetuses below 2000 g (36). Estimated fetal weight values were also dichotomized into low and normal fetal weight groups by determining whether estimated fetal weight values fell below the mean for a given gestational age.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
|
Fetal Variable Relationships and Analyses
Zero-order Pearson correlations revealed that fetal ultrasound biometry measures of biparietal diameter, head circumference, femur length, and abdominal circumference were significantly correlated (Table 1). Fetal biometry measurements were also significantly related to estimated fetal weight (Table 1). Correlation analyses also revealed that gestational age was significantly correlated to estimated fetal weight (r (98) = 0.96, p < .001) and fetal biparietal diameter (r (98) = 0.98, p < .001), head circumference (r (98) = 0.98, p < .001), femur length (r (98) = 0.95, p < .001), and abdominal circumference (r (98) = 0.97, p < .001).
Maternal/Fetal Variable Relationships and Analyses
Partial correlations, controlling for gestational age, were computed to examine whether maternal variables were associated with fetal measures. Partial correlations rather than zero-order correlations were used given the cross-sectional design of our study and the strong relation between gestational age and fetal growth. These analyses revealed the following (Table 1): a) maternal SES and age were not related to fetal biometry measurements; b) maternal depression scores were negatively related to biparietal diameter, head circumference, and fetal weight; c) maternal anxiety scores were negatively related to biparietal diameter, head circumference, and abdominal circumference; d) maternal hassles scores were negatively related to biparietal diameter; e) maternal cortisol was negatively related to biparietal diameter, head circumference, abdominal circumference, and fetal weight; and f) maternal norepinephrine was negatively related to biparietal diameter, head circumference, abdominal circumference, and fetal weight. Maternal weight at the time of the assessments was not related to gestational age or any fetal measurement.
Structural Equation Model
A structural equation model (SEM) was then conducted using structural equation modeling software (37) to assess whether maternal psychological distress and prenatal cortisol and norepinephrine were related to fetal weight while controlling for the effects of sociodemographic (ethnicity, marital status, age, and SES) variables on maternal psychological distress, and the effects of fetal gender and gestational age on estimated fetal weight. Fetal gender was dummy-coded (0 = female, 1 = male) and entered into the model. Inasmuch as maternal weight at the time of the ultrasound assessment was not related to any maternal or fetal variable, it was not included in the model. Estimated fetal weight obtained using the Shepard formula was the main dependent variable. Estimated fetal weight rather than individual ultrasound biometry measurements was used to reduce estimation errors (33). Analyses were conducted using the maximum likelihood estimation procedure. The hypothesized model was tested, and the comparative fit index (CFI) was used to evaluate the hypothesized model (Fig. 1). In this model, circles represent two latent variables (sociodemographics, psychological distress), and rectangles represent observed variables. Lines indicate a direct effect between two variables.
|
There was no missing data. All of the variables met assumptions for univariate and multivariate normality. Univariate and multivariate outlier inspection failed to reveal any out-of-range numbers. Pearson’s r computed for the composite scores obtained from normalized maternal social environment and psychological distress observed variables revealed a substantially smaller relation (r = 0.07) than the path obtained between the maternal social environment and psychological distress latent variables (r = 0.51), suggesting an adequate fit for the latent aspect of the model. The path between the maternal social environment and psychological distress further revealed that maternal sociodemographic characteristics significantly accounted for 26% of the variance in maternal psychological distress (Fig. 1).
A test of the hypothesized structural equation model using the maximum likelihood estimation procedure revealed a significant 2 (66, N = 98 = 64.79, p = .04), a CFI of 0.96, and a root mean square error of approximation (RMSEA) of 0.06, indicating an inadequate fit for the hypothesized model (Fig. 1). Post hoc modifications were then performed using the Wald and Lagrange Multiplier tests in an attempt to develop a better fitting and simpler model. The Wald test indicated that eliminating the paths between gestational age and maternal biochemistry and the paths between estimated fetal weight and gender, norepinephrine, and maternal distress would improve the fit of the model. As such, in an effort to attain parsimony, these paths were removed. Lagrange Multiplier tests failed to indicate any additional logical paths that would improve the fit of the model. The model was reestimated, revealing a significantly better fit for this model (2 (66, N = 98) = 42.46, p = .28, CFI = 0.99, RMSEA = 0.03; Fig. 2).
|
The direct and indirect effects of psychological distress on estimated fetal weight were then estimated by dropping the path between cortisol and fetal weight and adding a path between maternal distress and fetal weight. These analyses revealed that the path between psychological distress and fetal weight (Fig. 2, dashed path) was significant only when cortisol was not entered into the model (B = –0.08, p < .05, versus B = –0.04, not significant).
Odds Ratios
Odds ratios and confidence intervals were calculated to provide an estimate of the risk for low fetal weight as a result of elevated maternal cortisol. These analyses revealed that fetuses of mothers with elevated maternal cortisol levels were at risk for having fetuses with below average estimated fetal weight (odds ratio = 12.81; confidence interval = 4.81–34.09).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
The structural equation model revealed that maternal sociodemographic characteristics significantly accounted for 26% of the variance in maternal psychological distress (Fig. 2). This is not surprising given the strong link between psychosocial variables and maternal distress during pregnancy (40). Maternal daily hassles, anxiety, and depression scores were significantly related in our sample. This is consistent with research showing that hassles are a principal component of depression and anxiety (41) and the high comorbidity observed between depression and anxiety (42).
As predicted, maternal psychological distress (daily hassles, depression, and anxiety) was significantly related to both maternal urinary cortisol and norepinephrine levels. These findings are consistent with research indicating that maternal psychological distress during pregnancy is associated with both HPA and sympathoadrenal dysregulation (4,6,9,10,14–16). Maternal psychological distress was also significantly related to fetal biometry measures. These findings are consistent with our hypotheses and with research relating lower birthweight and prematurity to prenatal depression (4,12), anxiety (8), and stress (9,11).
Maternal urinary cortisol and norepinephrine levels were significantly related to each other (r = 0.51), which is not surprising given the common neural pathway linking the HPA and sympathetic nervous system. Activation of this pathway involves a cascade of events starting with the release of CRH from the hypothalamus and culminating in the release of cortisol from the adrenal cortex and norepinephrine and epinephrine from the adrenal medulla (39,43).
Both cortisol and norepinephrine were related to fetal weight; however, when the independent variance of maternal sociodemographic, psychological distress, and biochemistry measures were accounted for, prenatal cortisol was the only significant predictor of fetal weight. Cortisol and norepinephrine exhibit distinct molecular compositions and differ in their physiological effects. Cortisol readily crosses the placenta (44,45), whereas norepinephrine has not been shown to cross the placenta (46). Unbound cortisol inside the placenta can then affect fetal development by dysregulating placental CRH levels (47,48), inducing uterine artery contractions (49) and directly crossing into the fetus (48). Although these findings should be interpreted with caution given the small sample sizes at each gestational week, they tentatively suggest that elevated cortisol levels during midpregnancy may influence fetal growth.
The use of a cross-sectional sample with small numbers at each gestational week is perhaps the greatest limitation of this study. Examining fetal weight at a single time point would avoid dealing with the large proportion of the variance accounted for by gestation. Alternatively, a longitudinal sample allows for the study developmental trends. However, both of these methods would be difficult given the variability in time of prescribed ultrasounds and the reluctance to perform multiple ultrasounds because of potential side effects. Because we were unable to assess birthweight in this study, we could not assess fetal growth across the last trimester of pregnancy. Unfortunately, we also could not rule out the effects of maternal diet, vitamin use, or weight gain during pregnancy on fetal biometry measures, because these variables were not recorded in the present study. Future research should document those factors known to affect fetal growth.
Another potential limitation of the study is the reliability of the mothers reporting substance use and psychiatric conditions during pregnancy and the choice of measures used to assess social and psychological effects. For example, it is possible that the social stigma associated with substance use and psychiatric disorders during pregnancy led to underreporting these conditions. Similarly, although the CES-D and State-Trait Anxiety Inventory are reliable and have been extensively validated, they are questionnaires that are subject to self-report bias. In the present study, we measured stress using a self-report questionnaire of daily hassles recognizing that stress has not been consistently defined or measured. Finally, because we administered the self-report measures to the mothers at only one time point during pregnancy, we do not know if the depression, anxiety, and stress symptoms reported by the mothers varied in intensity or persisted throughout pregnancy.
In conclusion, the present study supports a pathway in which maternal psychological distress is accompanied by HPA dysregulation marked by elevated cortisol. Elevated cortisol may affect fetal growth through at least two potential pathways. The first pathway may involve maternal cortisol directly crossing the placenta and entering the fetus, resulting in a fetal HPA response (48). Fetal exposure to elevated cortisol, directly derived from the mother and indirectly derived from fetal HPA hyperactivation, may then affect fetal growth, increasing fetal calorie expenditure through increased neuromuscular activity and glycogenolysis (the conversion of glycogen to glucose) (50,51). The second pathway may involve maternal cortisol crossing the placenta and affecting the regulation of placental CRH (47,48), leading to placental hyperperfusion and resulting in reduced blood flow to the fetus (49). This reduction in blood flow to the fetus may in turn affect fetal growth by restricting oxygen and nutrient delivery (50,51).
Further work is needed to elucidate the underlying mechanisms for the effects of maternal psychological distress on fetal growth. Future studies could examine measures of maternal uterine artery resistance and measures of fetal activity and autonomic nervous system regulation to determine if maternal cortisol affects fetal growth through its effects on placental hyperperfusion and fetal energy expenditure.
This manuscript is based on a doctoral dissertation by Miguel A. Diego, PhD. We thank the women who participated in this study.
|
NOTES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Received for publication July 8, 2005; revision received May 18, 2006.
DOI:10.1097/01.psy.0000238212.21598.7b
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  I. J. Andersson, Y.-Y. Jiang, and S. T. Davidge Maternal stress and development of atherosclerosis in the adult apolipoprotein E-deficient mouse offspring Am J Physiol Regulatory Integrative Comp Physiol, March 1, 2009; 296(3): R663 - R671. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Li, L. Liu, and R. Odouli Presence of depressive symptoms during early pregnancy and the risk of preterm delivery: a prospective cohort study Hum. Reprod., January 1, 2009; 24(1): 146 - 153. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Maternal Stress and Midgestational Fetal Weight Journal Watch Women's Health, December 7, 2006; 2006(1207): 2 - 2. [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
