Prenatal Superstorm Sandy stress and postnatal affectionless control as a conjoint risk for child psychopathology

Abstract

The current study investigates the conjoint effect of prenatal Superstorm Sandy stress (SS) and postnatal parenting, as measured by affectionless control (AC), in determining the risk of early childhood psychopathology. The study included 154 children (45.5% males) longitudinally tracked from ages 2–5 years. Maternal AC was assessed using the Parental Bonding Instrument. The prevalence of child diagnostic outcomes (DSM-IV anxiety disorders, phobias, and disruptive behavior disorders) was ascertained by trained clinical interviewers using maternal responses to the Preschool Age Psychiatric Assessment. Children were stratified into four groups by SS and AC status to identify synergistic effects on psychopathology exceeding the risks expected in an additive model. Children exposed to both SS and AC had over 5-fold increased risk of any anxiety disorder, a 12-fold increased risk for disruptive behavior disorders, and a nearly 5-fold increased risk of any disorder relative to the reference group of children with neither exposure. The risks of anxiety, disruptive behavior, and any disorders were synergistically greater than the sum of independent effects of the two stressors, as evident in the synergy index. Evaluation of synergistically increased risks for childhood disorders will help to identify high-risk children, which in turn could inform design of multi-level interventions to mitigate child psychopathology.

INTRODUCTION

Prenatal stress negatively impacts endocrine and neurodevelopmental outcomes in offspring (Glover, 2011) through long-term perturbation of the hypothalamic-pituitary-adrenal (HPA) axis (Entringer et al., 2009, van den Bergh et al., 2020) and is associated with an increased risk for child psychopathology (de Bruijn et al., 2009). Childhood outcomes following fetal exposure to natural disaster-related stress are particularly important to understand given the increasing frequency and intensity of these events (van Aalst, 2006). Several longitudinal cohort studies have followed children born following prenatal exposure to ice storms, floods, hurricanes, and earthquakes, and show their heightened risk for poorer physical (e.g., Simcock et al., 2019), cognitive (e.g., Laplante et al., 2008), and psychiatric (e.g., Nomura et al., 2023) outcomes compared to non-exposed or Low-exposure peers.

Although the prenatal environment significantly shapes the developing fetus, a multitude of postnatal environmental factors moderate its influence on development to varying degrees over the life course (Grant et al., 2010; Kaplan et al., 2008). One potent postnatal environmental factor is parenting (Tarullo & Gunnar, 2006). In a double-hit acceleration model, the first exposure confers a risk or vulnerability for the second exposure to bring about a typically unfavorable outcome (Rothman, 2007; Saracci, 1980). Consistent with this model, the “first hit,” prenatal disaster-related stress, may alter the functioning of the HPA axis. When an individual subsequently encounters a “second hit” in the postnatal period – in this model, suboptimal parenting - its impact may be substantially greater than the single hit. That is, a conjoint risk of prenatal stress and suboptimal parenting may drive stress dysregulation and psychopathology (Koss & Gunnar, 2018).

An extensive body of literature demonstrates the importance of parenting in modulating the elevated risk of child psychopathology. Two broad dimensions have consistently been shown to emerge from descriptions of parents’ behaviors, characteristics, and interactions with their children; responsiveness and demandingness (Maccoby & Martin, 1983) or warmth/support and control (see Power, 2013, for a review). Two theoretical models have been influential in conceptualizing how parenting, characterized by differences along these two dimensions, impacts child developmental outcomes. Baumrind’s Y model (Baumrind, 1968) originally posited three parenting “styles” encompassing authoritative control (demanding, communicative, and nurturing), authoritarian control (demanding, but non-nurturing), and the lack of control (i.e., permissive control or non-controlling) (Baumrind, 1968). Maccoby and Martin (1983) reflected that parents showed different combinations of warmth/responsiveness and control/demandingness. They largely replicated and expanded on Baumrind’s work to propose four parenting “styles” that differed in degree of warmth and demandingness: authoritative (warm yet demanding), authoritarian (demanding but not warm), indulgent (warm but not demanding) and “uninvolved” or neglectful (low on both demandingness and warmth dimensions).

Baumrind’s work, and much more literature subsequently (e.g., Allmann et al., 2022), has shown that optimal parenting is high in warmth, but also sets expectations for and places limits around children’s behavior. A contrasting parenting approach is “affectionless control (AC),” characterized as lower on the warmth dimension and higher on the control dimension (Parker et al., 1979). Affectionless control is akin to Maccoby and Martin’s (1983) Authoritarian parenting style, which has been associated with greater risk for several forms of childhood psychopathology (e.g., Allmann et al., 2022; Valle et al., 2022). How parenting interacts with prenatal stress exposure to affect trajectories of children’s developmental outcomes is not yet known.

The current literature in this area has tended to operationalize prenatal stress as exposure to maternal psychopathology and examined its interaction with a specific parenting dimension, such as low warmth, sensitivity, or high control. The resultant findings suggest that prenatal exposure to stress coupled with suboptimal parenting confers an even greater risk for poorer outcomes than the presence of just one of these two risk factors. For example, low maternal sensitivity and warmth have been linked with high infant distress and negative affect, with highest levels of distress and negative affect observed in infants whose mothers had anxiety disorders during pregnancy (Grant et al., 2010). In another study, infants of mothers with prenatal anxiety disorders or depression had high cortisol levels only when their mothers were also low in sensitivity; for highly sensitive mothers, there was no notable difference in children’s baseline cortisol (Kaplan et al., 2008). In some qualitative case studies, parents with prenatal psychiatric diagnoses who had children with anxiety and depression were also found to engage in affectionless and overcontrolling parenting (Berg-Nielsen et al., 2002). Other studies investigated prenatal stress and parenting characteristics of mothers with a history of childhood trauma. Postnatal parenting that was less warm, more hostile and overcontrolling was linked with poorer behavioral and biological regulation in children. These children had lower levels of positive affect and heightened cortisol reactivity following a social stressor, maternal still-face (Martinez-Torteya et al., 2014). Likewise, for mothers with poor infant bonding, COVID-19 related prenatal stress, such as emotional distress and lack of social support, was inversely associated with regulatory ability of infants at 3 months (Provenzi et al., 2021). Deficits in self-regulation in infants are known to underlie and predict the expression of child psychopathology (Tronick & Beeghly, 2011; see Kostyrka-Allchorne et al., 2020 for a review).

Although the above human studies support the double-hit stress model, the operationalization of prenatal stress as preexisting psychiatric conditions is inherently confounded by factors that are difficult to eliminate (e.g., genetic susceptibility and comorbidities; Rice et al., 2010). The current study leverages a quasi-experiment of natural disaster-related prenatal stress. Superstorm Sandy hit the New York metropolitan area in October 2012, causing widespread disruption across the city, including loss of electricity, school shutdowns, gasoline shortages, and shutdown of and damage to the public transportation system (Blake et al., 2013). Based on maternal pregnancy status when the storm made landfall, maternal reports of the impact of the storm on their lives, and maternal perceived stress consequent to the storm, children were classified as having been exposed to Low or High levels of natural disaster-related stress in utero. We investigated the conjoint effect of prenatal exposure to Superstorm Sandy stress (SS) with postnatal maternal AC on the risk for psychopathology across early childhood. We hypothesized that children exposed to both High prenatal Superstorm Sandy stress and postnatal AC would have a substantially higher risk of psychopathology than those exposed to only one risk, and that these variables would interact synergistically to determine the risk profile.

METHODS

Participants

Pregnant women were recruited from the Obstetrics and Gynecology clinics at Mount Sinai Hospital and New York Presbyterian-Queens in New York, NY. Their babies were later born at these same hospitals between June 2011 and July 2015.

Inclusion criteria required the absence of obstetric risks such as HIV infection, maternal psychosis, maternal age <15 years old, life-threatening maternal medical complications, and congenital or chromosomal abnormalities in the fetus. The cohort encompassed an urban population with diverse ethnic/racial backgrounds and socioeconomic strata. Most participants identified as belonging to a racial/ethnic minority group and were socioeconomically disadvantaged. Following childbirth, child/mother dyads (N=358) were assessed annually. A subsample of 154 preschoolers with clinical interviews for preschool-aged children and assessments of parenting dimensions were included in the current study. Sixty-one (39.6%) children’s mothers were pregnant when Superstorm Sandy (SS) made landfall, of whom 23 (37.7%) were males.

Procedure

Mothers completed questionnaires about their parenting when children were 18, 24, 36, 48 and 60 months of age. Diagnostic outcomes prevalent in preschool aged children were assessed using a comprehensive structured battery of questionnaires administered by trained clinical interviewers to a subsample of 154 mothers with preschool-aged children. All participants provided written consent according to the protocol approved by the Institutional Review Board.

Measures

Prenatal stress.

Prenatal stress exposure status (“Superstorm Sandy Stress”) was defined according to whether mothers were pregnant and residing in the NY metropolitan area when SS made landfall in New York on October 29, 2012 (in-utero exposure) and the objective and subjective severity of stress that they experienced during and following Superstorm Sandy.

Objective Severity of Sandy Exposure was measured using Storm32 (King & Laplante, 2005). This questionnaire comprises 20 items that assess hardship experienced by individuals as a direct consequence of the storm (e.g., actual or threat of physical harm to self or loved ones, financial loss, property damage, loss of electricity or telecommunications, disruption to living situation). This measure has been shown to be reliable and valid in a large sample of mother-child dyads (Buthmann et al., 2019). Laplante and colleagues re-administered the measure to a subset of mothers who experienced the Quebec Ice Storm after a 6-year delay and obtained an overall intraclass correlation coefficient of ρ_I = .78, demonstrating good test-retest reliability (Laplante et al., 2008).

To create a binary variable of Low vs. High objective stress exposure, the number of stressors was summed. Women who experienced three or more stressors were classified as having had High exposure to storm-related stress, while those who experienced two or fewer were classified as having had Low exposure to storm-related stress.

Perceived Severity of Sandy Exposure was measured using the Impact of Events-Revised (IES-R, Weiss & Marmar, 1997) scale. This 22-item measure was modified to ask mothers to rate their subjective stress in the week following Superstorm Sandy. Items broadly correspond to DSM-IV PTSD symptoms and are rated on a 5-point Likert scale from 0 (Not at All) to 4 (Extremely). Weiss and Marmar (1997) showed test-retest reliability of the IES-R to be ≥.89 across 6 months. This measure has also been shown to be reliable and valid in a large sample of mothers who experienced a natural disaster (Buthmann et al., 2019).

To create a binary variable of Low vs. High subjective stress exposure, individual item responses were summed. Scores <30 were classified as Low Perceived Stress and scores ≥30 were considered indicative of High Perceived Stress.

A single Superstorm Sandy Stress (SS) variable was created that integrated the above three elements of in-utero exposure, objective stress, and perceived stress. Participants were considered to have “Low” exposure to Superstorm Sandy Stress if (i) their mother was not pregnant when the storm made landfall, OR (ii) if their mother was pregnant during the storm, but her objective stress was low AND her perceived stress was low. Participants were considered to have “High” exposure to Superstorm Sandy Stress if (i) their mother was pregnant when the storm made landfall AND (ii) her objective stress was High OR her perceived stress was High.

Parenting dimensions.

The Parental Bonding Index (PBI) was administered at 18, 24, 36, 48, and 60 months of age. The PBI measures two parenting dimensions, Care and Overprotection (Parker et al., 1979). One end of the Care dimension reflects affection, empathy, and emotional warmth compared to emotional coldness, indifference, and neglect at the other end. Overprotection, in contrast, reflects control, intrusiveness, stifling of independence and coldness at one end of the dimension and respect for autonomy and encouragement of independence at the other (Parker et al., 1979). Mothers responded to each of the 25 items using a 4-point Likert scale. The PBI was adapted in this study to allow mothers to rate their own parenting practices.

Internal consistency for the two dimensions was excellent (α = 0.90 and 0.92) in this study at baseline (age 18 or 24). In the current study, the correlations of PBI ratings across time points in the study ranged from 0.76 to 0.79. Murphy and colleagues (2010) showed test-retest correlations of .68 across 2 years for each dimension of the PBI.

The two parenting dimensions were then dichotomized at suggested cut-off points (Care <27; Overprotection/Control >13.5) to construct a binary index of Affectionless Control (AC) for mothers (Parker et al., 1979). AC was characterized by low Care and high Control. Caregivers who self-reported a Low Care/High Control parenting style at any of the five follow-up periods was considered to show AC parenting.

Risk groups by AC and SS.

To examine the magnitude of the conjoint effect of prenatal SS and postnatal AC, children were divided into four groups: 1) exposed to neither AC nor High Superstorm Sandy Stress (reference; AC-SS-), 2) exposed to AC only (AC+SS-), 3) exposed to High Superstorm Sandy Stress only (AC-SS+), and 4) exposed to both AC and High Superstorm Sandy Stress (AC+SS+). Table 1 shows that there were no differences in major demographic characteristics among the four groups.

Table 1.

Demographics characteristics of children exposed to neither prenatal Superstorm Sandy (SS) nor postnatal affectionless control (AC), only AC, only SS, and both (n=154)

		Exposure Groups				Statistics	Effect Size
	Total Sample N=154	Neither n =99 (1)	AC only n = 21 (2)	SS only n = 27 (3)	Both n =7 (4)
Child Race, N (%)						X2(9) = 6.91, p = .65	Cramers V = .12
White	64 (41.6)	38 (38.4)	8 (38.1)	16 (59.3)	2 (28.6)
Black	29 (18.8)	20 (20.2)	4 (19.0)	3 (11.1)	2 (28.6)
Asian	13 (8.4)	10 (10.1)	2 (9.5)	0 (0.0)	1 (14.3)
multiracial, race not listed	48 (31.2)	31 (31.3)	7 (33.3)	8 (29.6)	2 (28.6)
Child Ethnicity, N (%)						X2(3) = 5.08, p = .17	Cramers V = .18
Non-Hispanic	64 (41.6)	44 (44.4)	7 (33.3)	8 (29.6)	5 (71.4)
Hispanic	90 (58.4)	55 (55.6)	14 (66.7)	19 (70.4)	2 (28.6)
Child Sex, N (%)						X2(3) = 2.00, p = .57	Cramers V = .11
Males	69 (44.8)	46 (46.5)	11 (52.4)	10 (37.0)	2 (28.6)
Females	85 (55.2)	53 (53.5)	10 (47.6)	17 (63.0)	5 (71.4)
Socioeconomic Status, N (%)						X2(6) = 5.00, p = .54	Cramers V = .13
High	35 (22.7)	20 (20.2)	3 (14.3)	10 (37.0)	2 (28.6)
Medium	65 (42.2)	43 (43.4)	10 (47.6)	10 (37.0)	2 (28.6)
Low	54 (35.1)	36 (36.4)	8 (38.1)	7 (25.9)	3 (42.9)
Marital Status, N (%)						X2(9) = 4.50, p = .88	Cramers V = .10
Married	75 (48.7)	46 (46.5)	11 (52.4)	13 (48.1)	5 (71.4)
Common Law Marriage	7 (4.5)	6 (6.1)	0 (0)	1 (3.7)	0 (0)
Single	66 (42.9)	44 (44.4)	9 (42.9)	11 (40.7)	2 (28.6)
Separated/Divorced	6 (3.9)	3 (3.0)	1 (4.8)	2 (7.4)	0 (0)
Maternal Education, N (%)						X2(15) = 12.75, p = .62	Cramers V = .17
≤ Some high school	9 (5.8)	6 (6.1)	3 (14.3)	0 (0.0)	0 (0)
High School/GED	23 (14.9)	15 (15.2)	3 (14.3)	3 (11.1)	2 (28.6)
Some College	46 (29.9)	30 (30.3)	8 (38.1)	6 (22.2)	2 (28.6)
Associate Degreea	34 (22.1)	21 (21.2)	5 (23.8)	7 (25.9)	1 (14.3)
Bachelor’s Degreeb	40 (26.0)	25 (25.3)	2 (9.5)	11 (40.7)	2 (28.6)
Graduate/Professional	2 (1.3)	2 (2.0)	0 (0)	0 (0)	0 (0)
Normative Stress, N (%)						X2(6) = 4.85, p = .56	Cramers V = .13
High	27 (17.5)	15 (15.2)	5 (23.8)	6 (22.2)	1 (14.3)
Medium	77 (50.0)	47 (47.5)	12 (57.1)	13 (48.1)	5 (71.4)
Low	50 (32.5)	37 (37.4)	4 (19.0)	8 (29.6)	1 (14.3)
Maternal Age, Mean (SD)	27.69 (5.98)	27.38 (5.80)	27.13 (5.71)	28.06 (6.53)	32.23 (6.38)	F(3,150) = 1.55, p= .20	ηp2 = .03
Parity, Mean (SD)	2.10 (1.63)	2.03 (1.51)	2.38 (2.09)	2.15 (1.59)	2.00 (2.16)	F(3,150) = 0.28, p = .84	ηp2 = .006
Objective Sandy Stress, Mean (SD)	2.87 (2.70)	2.38 (2.35)	1.57 (1.08)	5.30 (3.05)	4.29 (3.68)	F(3,150) = 13.09, p < .001, 1<3; 2< 3	ηp2 = .21
Perceived Sandy Stress Mean (SD)	7.26 (12.70)	6.22 (12.87)	4.81 (8.73)	8.44 (11.14)	24.71 (14.64)	F(3,150) = 5.39, p < .001, 1,2, 3< 4	ηp2 = .10

^a2-year college

^b4-year college

Clinical Psychopathology: The Preschool Age Psychiatric Assessment (PAPA).

The PAPA is a parent-reported diagnostic interview designed to assess child psychopathology in children aged 2 to 5 years old (Egger & Angold, 2004). In this study, interviewers were graduate students who received intensive training and supervision by licensed clinical psychologists. The PAPA identifies DSM-IV disorders, including anxiety disorders (i.e., separation anxiety, generalized anxiety disorder, post-traumatic stress disorder, and selective mutism), phobias (i.e., specific phobia, social phobia, and agoraphobia), and disruptive behavior disorders (DBDs; i.e., conduct disorder, oppositional defiant disorder, and attention-deficit/hyperactivity disorder, ADHD). Interviewers conducted a thorough and in-depth investigation of the symptoms, including frequency, duration of symptoms, and age of symptom onset. If a child’s mother endorsed a sufficient number of symptoms for each disorder according to the DSM-IV during this face-to-face clinical interview, the child was considered positive for the disorder. Results were then aggregated across diagnostic clusters. For example, for DBDs, if a child’s mother endorsed at least one of conduct disorder, oppositional defiant disorder, or ADHD, the children was considered positive for DBDs.

Egger and colleagues (2006) evaluated test-retest reliability of the PAPA, interviewing 114 caregivers twice over the period of approximately 1–2 weeks (mean=11 days). PAPA’s reliability for diagnostic categories was comparable to that of commonly used measures for older children and adults, κ= 0.36 (specific phobia) to κ=0.87 (Enuresis) (Egger et al., 2006). In the current study, Inter-rater reliabilities were fair to good for dysthymia (κ = .72), specific phobia (κ = .46), social phobia (κ = .54), separation anxiety disorder (κ = .60), generalized anxiety disorder (κ = .59), selective mutism (κ = .88), conduct disorder (κ = .66), oppositional defiant disorder (κ = .62), and attention-deficit/hyperactivity disorder (κ = .78).

Covariates

Child and Maternal Demographic Variables.

Mothers reported salient aspects of their own and their child’s identity, including child sex, race and ethnicity, mother and child birth dates, maternal highest level of education, marital status and parity. Maternal age at the time of child birth was calculated from mother and child dates of birth. Maternal education was used as the indicator of socioeconomic status (SES) in this study to reduce the potential for bias that can arise from other measures, such as employment or income, if mothers chose to stay at home during pregnancy.

Normative prenatal stress.

This was extracted using latent profile analysis (LPA, Tein, Coxe, & Cham, 2013) from the following measures: Pregnancy-Related Anxiety Questionnaire-Revised (PRAQ-R; Huizink et al., 2004), the 14-item Perceived Stress Scale (PSS-14) (Cohen & Williamson, 1988), Edinburgh Postnatal Depression Scale (EPDS) (Cox et al., 1987) and the State- and Trait- Anxiety Inventory (STAI) (Spielberger, 1989). All measures show adequate reliability and validity (Cohen & Williamson, 1988; Cox et al., 1987; Huizink et al., 2004; McDowell, 2006). LPA tested 2- and 3-class solutions, which both showed good model fit with entropy of 0.8 or greater and significant Lo-Mendell-Rubin (L-M-R) test scores. Both the BIC and AIC values decreased from the 2-class model to the 3-class model. When comparing the model fits between 2-class and 3-class models, there was a significant improvement (p=.002). Taken together, the 3-class model (low, medium, and high) was selected as our index.

Statistical methods

Descriptive analyses were conducted to examine demographic differences among the reference, AC only, SS only, and both AC and SS risk groups using analysis of variance for continuous variables and chi-square test for categorical variables.

Following descriptive analyses, survival analysis techniques were used to evaluate the differential age of first onset and cumulative risks of childhood diagnostic outcomes. First, a Wilcoxon test was used to examine the equality of survival distributions among children exposed to neither AC nor SS (reference), those exposed to AC only, those exposed to SS only, and those exposed to both AC and SS. Second, to evaluate the cumulative risk of developmental psychopathology, a Cox proportional hazards regression model (Cox, 1972) was used to estimate the hazard ratio (HR) of disorders in the three experimental groups relative to the reference group. Synergy index (SI) was calculated to estimate the magnitude of the interaction on an additive scale (Knol et al., 2011). SI refers to the ratio of the additive effect of risk associated with exposure to both AC and SS over the sum of the two single risks of AC and SS. SI was calculated with the formula, SI_AB = (HR_AB – 1)/(HR_A + HR_B – 2), where SI_AB and HR_AB refer to SI and HR of the conjoint effect of AC and SS respectively, and HR_A and HR_B refer to HR of individual effect of AC and SS respectively. An SI of 1 represents no additive interaction (i.e., no synergy) as evidenced by the risk of the conjoint risk of both AC and SS equals the sum of the single risks (Rothman, 2012). Generally, the conjoint risk or synergy is considered as appreciable when SI exceeds 1 (Rothman, 2012).

A priori determined covariates, including child sex, child race, child ethnicity, maternal age, maternal education level, marital status of the parents, parity, socioeconomic status (SES), and normative prenatal stress, were controlled in all models for statistical adjustment.

RESULTS

Demographic characteristics by prenatal SS and postnatal AC

There were 99 children in the reference group, 21 in the Affectionless Control (AC) only group, 27 in the Superstorm Sandy Stress (SS) only group, and 7 in both AC and SS group. As displayed in Table 1, there were no significant demographic differences among the four groups, including child race, child ethnicity, child sex, SES, marital status, maternal education, prenatal normative stress, maternal age, and parity.

Risk of psychopathological disorders in early childhood by prenatal SS and postnatal AC

The test of equality of strata showed a significant difference between the four groups, suggesting the time of the initial onset of child psychopathology differed among the four groups for anxiety disorders (χ²(3) = 16.83, p < .001), DBDs (χ²(3) = 11.20, p = .01) and any disorder (χ²(3) = 12.36, p = .006).

Table 2 shows the cumulative prevalence rates of child psychopathology of the four groups. Children exposed to both SS and AC had the highest prevalence for anxiety disorders (85.7%), phobias (71.4%), DBDs (42.9%) and any disorder (100.0%). Children exposed to SS only had the second highest cumulative risk in anxiety disorders, DBDs, and any disorder. The children exposed to AC only and the reference group had similar prevalence rates.

Table 2.

Risk of disorders among children with prenatal exposure to High Superstorm Sandy Stress (SS) and/or postnatal Affectionless Control (AC) over the first five years of life (n=154)

	Cumulative Rates over 5 years, N (%)				Overall difference	Hazard Ratio (HR)				Synergy Indexa (SI)
	Neither (n=99)	AC only (n=21)	SS only (n=27)	Both (n=7)	Wald X2(3), p	Neither (Reference)	AC only HR (95% CI), p-value	SS only HR (95% CI), p-value	Both SS and AC HR (95% CI), p-value
Anxiety Disordersb	26 (26.3)	5 (23.8)	17 (63.0)	6 (85.7)	16.83, p < .001	1.0	0.80 [0.29, 2.23], p = .67	2.51 [1.27, 4.96], p = .008	5.03 [1.90, 13.31], p =.001	3.08
Phobiasb	43 (43.4)	8 (38.1)	14 (51.9)	5 (71.4)	7.79, p = .05	1.0	0.74 [0.33, 1.68], p = .48	0.98 [0.51, 1.89], p = .96	4.04 [1.42, 11.52], p = .009	-e
DBDsc	14 (14.1)	2 (9.5)	7 (25.9)	3 (42.9)	11.20, p = .01	1.0	0.47 [0.10, 2.26], p = .35	1.63 [0.60, 4.43], p = .34	12.14 [2.49, 59.19], p = .002	111.40
Any disorderd	51 (51.5)	12 (57.1)	21 (77.8)	7 (100.0)	12.36, p = .006	1.0	1.07 [0.53, 2.15], p = .86	1.58 [0.91, 2.76], p = .11	4.57 [1.86, 11.25], p < .001	5.39

^aSynergy index (SI) was calculated as follows: SI_AB= (HR_AB - 1)/(HR_A + HR_B - 2)

^bAnxiety disorders do not include phobias and phobias does not include anxiety disorders

^cDBDs = disruptive behavior disorders

^dAny disorder includes any of the positive diagnosis of anxiety disorders, phobias, and disruptive behavior disorders.

^eNot estimable

Analysis was based on the proportional hazard model with covariates including child sex, child race, child ethnicity, maternal age, marital status of the parents, maternal education, parity, SES, and normative prenatal stress.

The second column shows the cumulative risk for children exposed to AC only, to SS only, and both AC and SS as compared to the reference group to test for a synergistic increase in risk induced by the joint exposure. Compared to the reference group, children exposed to both AC and SS had a 5-fold increase in anxiety disorders (HR = 5.03, 95% CI [1.90, 13.31], p = .001), a 12-fold increase in DBDs (HR = 12.14, 95% CI [2.49, 59.19], p=.002), and an over 4-fold increase in any disorder (HR = 4.57, 95% CI [1.86, 11.25], p < .001). Children exposed to SS only had an over 2-fold increase in anxiety disorder (HR = 2.51, 95% CI [1.27, 4.96], p = .008), but no difference in risk between the SS only group and the reference group was seen for disruptive behavior disorders, phobias or any disorder. The AC only group did not differ significantly from the reference group. Age at first onset of psychopathology is shown in Figure 1. Last, the conjoint effect of AC and SS showed synergistic increases in risk for anxiety disorder (SI = 3.80), disruptive behavior disorder (SI = 111.40) and any disorder (SI = 5.39) compared to the sum of the two risks.

Figure 1.

Survival curves for disorders among children with prenatal exposure to High Superstorm Sandy stress (SS) and/or postnatal Affectionless Control (AC) over the first five years of life (n=154).

Note. A= Anxiety Disorders; B=Phobias; C=Disruptive Behavior Disorders; D=Any Disorder

Legend: _____ SS – AC – __ . __ SS – AC + _ _ _ _ SS + AC – _____ SS + AC +

a. SS – = Low exposure to Superstorm Sandy Stress (mother not pregnant when Sandy made landfall or mother was pregnant, but experienced low objective storm-related stress and low perceived stress. SS + = High exposure to Superstorm Sandy Stress (mother pregnant when Sandy made landfall and mother experienced high objective storm-related stress or high perceived stress. AC - = parenting approach characterized by high care/low control, high care/high control, or low care/low control. AC + = parenting approach characterized by low care and high control.

b. Objective Superstorm Sandy-related stress measured using the Storm32. High objective stress operationalized as exposure to three or more stressors. Perceived Superstorm Sandy-related Stress measured using the Impact of Events Scale-Revised. High perceived stress operationalized as a total score at or above 30. Affectionless Control measured using the Parental Bonding Instrument. Low Care was operationalized as a score <27 and High Overprotection/Control was operationalized as a sore >13.5.

c. Covariates in all models included child sex, child race, child ethnicity, maternal age, marital status of the parents, maternal education, parity, SES, and normative prenatal stress.

d. A significant difference among groups was found in the test of equality of strata (Wilcoxon test), χ2(3) = 16.83, p < .001 for any anxiety disorder; χ2(3) = 11.20, p = .01 for any disruptive behavior disorder; χ2(3) = 12.56, p = .006 for any disorder, but not for phobias, χ2(3) = 7.79, p = .05.

DISCUSSION

We investigated the interaction of prenatal Superstorm Sandy Stress (SS) and postnatal maternal Affectionless Control (AC) in the risk for anxiety disorders, phobias and disruptive behavior disorders (DBDs) using the framework of the double-hit model. The greatest cumulative risk for anxiety disorders, DBDs, or any disorder, was found in children exposed to both AC and SS. Prenatal exposure to SS independently produced the second greatest risk for anxiety disorders. The synergy indices (SI) showed that the elevated risk of DBDs, anxiety disorders, or any disorder associated with joint exposure to both AC and SS was greater than the sum of the independent risks from the two exposures, providing initial evidence for the additive interaction, or synergy by AC and SS exposures.

Our findings support a double-hit model whereby children exposed to both prenatal Superstorm Sandy Stress and postnatal AC experience synergistically elevated risks of anxiety disorders, DBDs, and any disorder. This is consistent with prior findings that prenatal stress and postnatal parenting, characterized as hostile-reactive parenting, were associated with higher externalizing problems among children between 23 and 60 months of age (Hentges et al., 2019). In another study, prenatal maternal depression and anxiety were associated with more internalizing problems (i.e., depressive and anxiety symptoms) when maternal parenting qualities were less optimal (i.e., low sensitivity, high intrusiveness, and high hostility; Endendijk et al., 2017). Unlike our quasi-experimental prospective design, however, these studies were cross-sectional and non-quasi-experimental designs. In addition, these studies investigated a narrow subset of behavioral concerns, while this study used diagnostic outcomes ascertained by structured clinical interviews by clinicians who underwent intensive training, monitoring, and interrater reliability testing.

To our knowledge, there has been only one prior study that examined the effects of prenatal natural disaster (Queensland Flood) and postnatal parenting on children’s outcomes. The study found that prenatal maternal stress and postnatal maternal overinvolvement did not exacerbate anxiety symptoms of school-aged children at 6 years (McLean et al., 2021). In contrast our study found that the conjoint exposures to both prenatal disaster-related stress and postnatal parenting low in Care and high in Overprotection was associated with synergistic increases in risk for anxiety disorders, DBDs, or any disorder in preschool-aged children. The inconsistent findings between the two studies may be attributable to methodological differences and the differences between the parenting measured. First, the Queensland Flood study used the caregiver-reported Spence Preschool Anxiety Scale, while our study measured diagnostic outcomes using structured clinician-administered interviews. Second, the Queensland Flood study evaluated the conjoint effect of prenatal stress and postnatal parenting using multiplicative interaction, whereas our study used additive interaction (Rothman, 2012). We chose additive interaction because it enabled us to evaluate the degree of synergy between the two stressors in determining the risk of psychopathology (VanderWeele & Robins, 2007). Last, while the parenting dimensions studied in the Queensland Flood study and the current study share some features, they have important differences. The Queensland Flood study only assessed Overinvolved parenting (McLean et al., 2021), where the current study assessed maternal AC parenting characterized by low levels of care and high levels of control/overprotection. Overinvolved parenting does not speak to the level of parental warmth, which may be driving these differences. Our findings should be replicated and extended to gain further knowledge about the combined exposures, and impact of timing of exposures (prenatal and postnatal), on the risk of developmental psychopathology. This is particularly important given the increasing frequency and intensity of natural disasters (van Aalst, 2006).

Additive interaction has been suggested to evaluate biological interaction, such as HPA-axis biological mechanism (Knol, et al., 2011). However, the biological underpinnings of our findings are not fully understood. It is possible that the conjoint effect of prenatal and postnatal insults operates through altered neuroendocrine mechanisms to modulate child psychopathology risk. Fetal reprogramming of HPA axis has been widely proposed as one of the biological mechanisms of transmission of prenatal stress to the child (Sloboda et al., 2006). Specifically, elevated levels of glucocorticoids from the mother can be transmitted to the fetus through the placenta and upregulate the stress-response system of offspring to increase their chance of survival in a stressful environment after birth. Children of mothers who experienced intimate partner violence during pregnancy have elevated cortisol secretion, which is linked to child internalizing problems such as major depression (Martinez-Torteya et al., 2016). Similarly, postnatal overcontrolling parenting has been shown to influence the HPA axis and child psychopathology (Tarullo & Gunnar, 2006); parent-child co-regulation may place an important role on stress regulation of the child (Gunner & Quevedo, 2007). A mother who is overcontrolling and affectionless toward the child may fail to provide the necessary scaffolding of emotional regulation for the child as they learn to process and respond to their own emotions.

Other research has shown that children whose mothers were exposed to prenatal stress and parented with AC had higher levels of cortisol production (Kaplan et al., 2008; Martinez-Torteya et al., 2014), as compared to the reference group that were exposed to neither. Consistent with associations among prenatal stress, postnatal AC, and HPA axis, it has been found that the impact of prenatal stress and postnatal AC on elevated internalizing problems during childhood was mediated through morning cortisol level of the offspring (Marceau et al., 2015). Recent neuroimaging studies further support possible HPA-axis-related biological mechanisms. Reduction in size of the hippocampus, an area rich in glucocorticoid receptors, is implicated in development of psychiatric disorders (Gilbertson et al., 2002; Rao et al., 2010). Indeed, low birth weight (a proxy for higher prenatal stress) predicted smaller adult hippocampus volume among females parented with postnatal AC (Buss et al., 2007). Taken together, those studies highlight the importance of revisiting additive interaction (i.e., synergy between the two risk factors) to understand health outcomes.

The current study suggests that independent prenatal exposure may have a larger impact on child psychopathology than parenting. One explanation for this may be rooted in the fetus being directly exposed to maternal stress hormones through the placenta. The placenta plays a pivotal role in providing nutrients from the mother to the fetus, but it can also serve as a bridge to directly transfer high concentrations of signaling molecules associated with prenatal stress, leading to neurodevelopmental perturbations (Monk et al., 2020). The results from the current study may suggest that stress related to prenatal exposure to natural disasters may directly affect fetal brain development, which can increase the risk for later child psychopathology. However, it is important to note that exposure to both prenatal Superstorm Sandy related stress and postnatal AC was associated with the greatest increased risk in childhood anxiety and DBDs. Future studies are needed to clarify the underlying connections and understand further why postnatal suboptimal parenting had greater implications for children who were prenatally exposed to disaster-related stress than those were not.

Implications of findings

The results highlight two critical targets for intervention aimed at reducing risk of child psychopathology. Health care professionals should be cognizant of the impact of stress on the developing fetus and be able to communicate this in plain language to pregnant women. Interventions and strategies that reduce stress in pregnant women should be promoted not only at prenatal appointments, but among community resources so they are available to women irrespective of access to health care. Natural disasters are occurring more frequently and with greater intensity given climate change (van Aalst, 2006). For a pregnant woman, these events pose challenges as food security, shelter, access to health care, and sanitation are vital for her own health and that of her fetus. Worry may compound difficulties, and resultant stress increases risk for poorer perinatal and childhood outcomes. Disaster planning by local and national governments, as well as non-governmental organizations, should consider resources for women who are pregnant and how to efficiently distribute them in times of crisis (Ha, 2022; Nour, 2011).

Furthermore, given the additive impact of maternal post-natal parenting on childhood outcomes, parent training interventions present a promising avenue for reducing the risk of child psychopathology. Parenting interventions have been shown to improve early child outcomes (e.g., cognition, language), child attachment, socioemotional functioning, and reduce behavior problems. These benefits were amplified when the intervention incorporated a component to target responsive parenting (Jeong et al., 2021). Educators and health care professionals need to be mindful of potential barriers to access interventions and strive to implement these interventions in an equitable manner across diverse communities. Many studies have delivered the intervention in people’s homes, which can enhance engagement and provide learning opportunities in a natural environment (Jeong et al., 2021). There is evidence to suggest cross-cultural applicability of parenting interventions (Maciel et al., 2023) as well as effectiveness in low- and middle-income countries (Jeong et al., 2021).

Strengths and limitations

The study has several strengths. First, we leveraged natural disaster exposure in a quasi-experimental paradigm to control for variables that may confound prenatal stress, such as genetic predisposition and maternal psychopathology. Second, the study offers greater accuracy in the diagnosis of child psychopathology through structured, clinician-administered face-to-face interviews. Third, studies on child psychopathology in preschool aged children have been underrepresented in the literature on prenatal stress and parenting behaviors, attitudes, and practices, so the present study fills an important knowledge gap. Last, the study’s longitudinal prospective design strengthens our assessment of causal relationships between variables.

The study also has limitations. First, clinical interviews were based solely on maternal report. Maternal psychopathology such as depression and anxiety may influence how mothers report the developmental and behavioral patterns of their children, although a recent study demonstrated no evidence for such bias (Olino et al., 2021). We did not include a multi-informant rated measure of parenting. The sole use of mother-reported parenting may have left room for bias in mothers’ perception of how they relate to their children, as well as level of and methods used to control children’s behavior. Second, parenting was assessed by a static measure of AC, which cannot give information about the reciprocal interaction of parenting and child emotional and behavioral difficulties (e.g., Smith et al., 2014).

Additionally, parenting is culture-dependent, and outcomes may vary accordingly. Most studies about parenting have been conducted in the United States or United Kingdom with predominantly White, middle-class families (Lansford, 2022), and these have tended to show that authoritative parenting style (i.e., warmth and strictness) is associated with the best psychosocial adjustment of the child (Steinberg et al., 1994). However, parenting must be considered within the unique cultural context in which it is occurring as this affects both its form and function (Lansford, 2022). Indulgent parenting (i.e., parental warmth without parental strictness) has been associated with better child outcomes in European and Latin American countries (e.g., Fuentes et al., 2022; Martinez et al., 2020). In contrast, Authoritarian parenting has been associated with more positive outcomes among some Arab (Dwairy & Achoui, 2006), Black and African American (Deater-Deckard et al., 1996), and Chinese American families (Chao, 2001). Third, we did not assess the role of paternal parenting in modulating the risk of psychopathology. Fourth, we used the dichotomous predictor of High levels of Superstorm Sandy-related stress in utero without accounting for trimester-specific timing of the exposure. Fifth, we controlled various a priori sociodemographic confounders, but other confounds might have been overlooked, such as time confound. Sixth, we administered the PAPA diagnostic interview multiple times between age 2 and 5 to minimize the retrospective recall error. However, we still cannot fully rule out recall bias on age of the first onset via the PAPA diagnostic interviews. Last, specific psychopathological disorders may be considered to belong under different classes than utilized in this study (e.g., PTSD classified as an anxiety disorder; APA, 2000, 2013). Hence, the magnitude of group differences may be impacted by inclusion or exclusion of certain disorders within each class.

Future directions

Future research should seek to understand the impact of cultural differences on parenting, and how this modulates the effects of prenatal exposure to disaster-related stress. Studies should also examine the efficacy and effectiveness of parenting interventions on children’s behavior and physiological responsivity, and the mechanism of action (Gunnar, 2023).

CONCLUSIONS

This longitudinal quasi-experimental study provides support for a double-hit stress model where the conjoint effect of prenatal Superstorm Sandy (SS) and postnatal Affectionless Control (AC) produces synergistically elevated risk for anxiety disorders and disruptive behavior disorders among preschool-aged children. Specifically, the first hit of prenatal SS stress enhances vulnerability to these particular internalizing and externalizing psychopathologies and accelerates the negative impact of the second hit of postnatal suboptimal parenting. Consistent with the double-hit hypothesis, our results showed that the dual exposure of prenatal SS and postnatal AC was associated with more than a 5-fold increase in the risk of anxiety disorders, a 12-fold increase in any disruptive behavior disorders, and a 4-fold increase in any disorder. It should be reminded that the double hit of prenatal SS and postnatal AC also synergistically increased the risk of anxiety disorders and DBDs. Specifically, the conjoint risk of AC and SS was 3-fold greater than the sum of individual risks from AC and SS for anxiety disorders, over 100-fold greater for DBDs, and 5-fold greater for any disorder. These findings indicate that once exposed to prenatal natural disaster stress, the impact of subsequent negative influence on the risk of childhood psychopathology was substantially and disproportionally elevated. Nature and nurture forces conjointly elevate the risk of child psychopathology and engender long standing effects on the developing child. During the critical period of early childhood development, the manner in which a parent sets expectation for child behavior and the degree of emotional support and responsiveness to their child’s needs is important. High expectations for obedience, high structure and low levels of warmth amplify biological risk conferred during the prenatal period to increase risk for later anxiety and disruptive behavior problems. The adverse effects of certain dimensions of parenting on child mental and behavioral health is not new. However, given the increased incidence of natural disaster in recent years, it is important for educators, policy makers and health care providers to become aware that parenting could have enormous implications on the well-being of our young children. Early identification of children in this high-risk double-hit group and multi-level interventions from the prenatal to postnatal period are warranted.

Highlights:

• This longitudinal study tracks 154 children between 2 and 5 years of age.

• Children were stratified into four groups by prenatal Superstorm Sandy stress and postnatal affectionless control. Synergistic, co-joint effect was greater than the sum of the two stressors.

• Children with double hits were high-risk for anxiety, DBD, and any disorders.