Received: 12 October 2021 | Revised: 24 April 2022 | Accepted: 17 June 2022

DOI: 10.1002/da.23277

RESEARCH ARTICLE

Contribution of early‐life unpredictability to neuropsychiatric

symptom patterns in adulthood

Andrea D. Spadoni1,2,3 | Meghan Vinograd2,3 | Bruna Cuccurazzu2,3 |

2,3 4,5 6,7
Katy Torres | Laura M. Glynn | Elysia P. Davis | Tallie Z. Baram7,8,9 |
Dewleen G. Baker2,3 | Caroline M. Nievergelt2,3 | Victoria B. Risbrough2,1,3

1
VA San Diego Healthcare System, San Diego,
California, USA Abstract
2
Department of Psychiatry, University of Background: Recent studies in both human and experimental animals have identified
California, San Diego, San Diego,
California, USA
fragmented and unpredictable parental and environmental signals as a novel source
3
Center of Excellence for Stress and Mental of early‐life adversity. Early‐life unpredictability may be a fundamental developmen-
Health, VA San Diego Healthcare System, San tal factor that impacts brain development, including reward and emotional memory
Diego, California, USA
4
circuits, affecting the risk for psychopathology later in life. Here, we tested the
Department of Psychology, Chapman
University, Orange, California, USA hypothesis that self‐reported early‐life unpredictability is associated with psychiatric
5
Department of Psychiatry and Human symptoms in adult clinical populations.
Behavior, University of California, Irvine,
Methods: Using the newly validated Questionnaire of Unpredictability in Childhood,
Irvine, California, USA
6 we assessed early‐life unpredictability in 156 trauma‐exposed adults, of which 65%
Department of Psychology, University of
Denver, Denver, Colorado, USA sought treatment for mood, anxiety, and/or posttraumatic stress disorder (PTSD)
7
Department of Pediatrics, University of symptoms. All participants completed symptom measures of PTSD, depression and
California, Irvine, Irvine, California, USA
anhedonia, anxiety, alcohol use, and chronic pain. Relative contributions of early‐life
8
Department of Anatomy/Neurobiology,
University of California, Irvine, Irvine, unpredictability versus childhood trauma and associations with longitudinal
California, USA outcomes over a 6‐month period were determined.
9
Department of Neurology, University of Results: Early‐life unpredictability, independent of childhood trauma, was signifi-
California, Irvine, Irvine, California, USA
cantly associated with higher depression, anxiety symptoms, and anhedonia, and was
Correspondence related to higher overall symptom ratings across time. Early‐life unpredictability was
Victoria B. Risbrough, Department of
also associated with suicidal ideation, but not alcohol use or pain symptoms.
Psychiatry, University of California, San Diego,
9500 Gilman Dr., La Jolla, CA 92093‐0804, Conclusions: Early‐life unpredictability is an independent and consistent predictor of
USA.
specific adult psychiatric symptoms, providing impetus for studying mechanisms of
Email: [email protected]
its effects on the developing brain that promote risk for psychopathology.
Funding information
Veteran Affairs, Grant/Award Numbers: KEYWORDS
5I01BX004312, 5I01CX001762; National
anhedonia, anxiety, childhood trauma, depression, early‐life adversity, posttraumatic stress,
Institute of Mental Health,
unpredictability
Grant/Award Number: P50MH096889

1 | INTRODUCTION nutritional deficits are well‐established moderators of brain circuit

development and subsequent behavioral functions (Frewen et al., 2012;
The developing brain is highly sensitive to environmental signals, many of Pechtel & Pizzagalli, 2013). Childhood adversity and trauma are
which have long‐term effects on adult traits and neuropsychiatric risk. associated with increased risk for mood and anxiety disorders in addition
Environmental factors such as trauma, socioeconomic disadvantage, and to physical health problems in adulthood (Agorastos et al., 2014). Recently

Depression and Anxiety. 2022;1–12. wileyonlinelibrary.com/journal/da © 2022 Wiley Periodicals LLC. | 1

2 | SPADONI ET AL.

proposed theoretical models suggest that differential forms of adversity environment may tap into a fundamental developmental factor that
may have distinct effects on outcomes relevant to psychopathology shapes neural development and risk for psychopathology in
(McLaughlin & Sheridan, 2016). A more comprehensive understanding of adulthood (Davis et al., 2019; Doom et al., 2016; Glynn & Baram,
the neuropsychiatric outcomes associated with particular forms of early‐ 2019; Glynn et al., 2019; Granger et al., 2021; McGinnis et al., 2022;
life adversity may aid in prevention and intervention efforts. Risbrough et al., 2018). To date, most studies have used samples that
Integrated studies in both experimental systems and humans were not selected for psychiatric risk or current symptoms.
have identified a novel source of developmental effects on neural and Therefore, there is little information about how early‐life
behavioral functions: fragmented and unpredictable signals, particu- unpredictability contributes to symptoms in psychiatric populations.
larly those derived from maternal care and the home environment Filling this gap is an important step in understanding what dimensions
(Davis et al., 2017; Glynn et al., 2018; Molet, Heins et al., 2016; of early‐life adversity contribute to neuropsychiatric risk, and if these
Risbrough et al., 2018). Fragmented and unpredictable patterns of dimensions individually contribute to distinct or specific symptom
sensory signals from the mother were originally described in animal patterns.
models of simulated early‐life poverty. In these studies, observation Here, we tested the hypothesis that early‐life unpredictability
and quantitative measures of maternal care demonstrated that while explains unique variance in symptom domain and severity in adults
the quantity and typical qualitative measures of care were provided, with elevated mood and anxiety symptoms. We utilized a well‐
the patterns of care behaviors (i.e., the sensory and tactile signals) validated self‐report measure of early‐life unpredictability, the
received by the pups were disrupted via stress experienced by the Questionnaire of Unpredictability in Childhood (QUIC; Glynn et al.,
mother (Chen & Baram, 2016; Ivy et al., 2008; Molet et al., 2016). 2019), and established clinical measures to assess psychiatric
These studies revealed that, in rodents, unpredictable care disrupts symptoms. To test specificity, we examined associations of early‐
reward circuits and reward‐seeking behaviors and predicts life unpredictability with alcohol abuse and pain symptoms. To
anhedonia‐like behaviors (Bolton et al., 2017, 2018; Molet, understand the relative contributions of different forms of early‐life
Heins et al., 2016). adversity, we compared the effects of early‐life unpredictability with
In humans, fragmented and unpredictable early‐life experiences the effects of childhood abuse and neglect, as measured with the
are measured across multiple domains, including maternal sensory Childhood Trauma Questionnaire (CTQ; Bernstein et al., 1994).
signals, maternal mood, and caregiving and environmental experi- Finally, we tested the hypothesis that these associations may predict
ences during childhood and adolescence. Exposure to higher symptom change over time, by examining associations of QUIC
unpredictability of maternal sensory signals in the first year of life scores, time, and their interaction on symptoms assessed 3‐ and 6‐
predicts lower cognitive and language skills, poorer performance on a month intervals.
hippocampus‐dependent memory task, and poorer effortful control
across early and middle childhood (Davis et al., 2019, 2017). Further-
more, higher unpredictability of maternal mood during the prenatal 2 | METHODS
period predicts lower cognitive development, lower expressive
language, higher levels of child negative affectivity, and higher self‐ 2.1 | Participants
reported anxiety and depressive symptoms across childhood
and early adolescence (Glynn et al., 2018; Howland et al., 2020). Participants (N = 156 at intake assessment, mean age = 43.03 years,
Self‐reported exposure to unpredictability in social, emotional, and 62% male) seeking treatment at Veterans Affairs (VA) clinics or
physical domains during childhood and adolescence predicts lower interested in mental health research studies at the VA were referred
emotional control, as well as greater symptoms of anxiety, depres- by staff or were recruited through flyers and word of mouth to
sion, and anhedonia in adolescents and adults (Glynn et al., 2019; participate in the Center of Excellence for Stress and Mental Health
McGinnis et al., 2022; Szepsenwol et al., 2021). Exposure to early‐life (CESAMH) posttraumatic stress disorder (PTSD)/Traumatic Brain
unpredictability is also associated with an increased risk of later Injury(TBI) Biorepository Study at the San Diego Veterans Affairs
substance use, conduct problems, risk‐taking behavior, and poor Healthcare System (SDVAHS). Mental health symptoms were
relationship quality (Doom et al., 2016; McGinnis et al., 2022; assessed via standardized self‐report measures presented over an
Simpson et al., 2012; Szepsenwol et al., 2021). Taken together, electronic tablet (eScreening; Pittman et al., 2017), as well as a brief
unpredictability across domains and age of exposure are associated interview with a research assistant to assess current treatment
with important cognitive, biological, affective, and social outcomes status. Participants provided tissue for banking (blood, urine, and
across the lifespan. saliva; tissue data not included in this analysis). Participants were
Importantly, many of these findings were in cohorts with assessed up to three times, with approximately 3 months between
relatively low trauma exposure, suggesting potential unique effects assessments to measure potential symptom changes over a 6‐month
of early‐life unpredictability in childhood. Contributions of early‐life period. Participants were recruited through behavioral health clinics
unpredictability on emotional and cognitive functions have now been or through collaborating mental‐health treatment studies at the
replicated across laboratories and human cohorts internationally, SDVAHS. The majority (N = 130) of the study population had past
suggesting that unpredictable signals from caregivers and the military service or were Veterans seeking treatment at the VA for
SPADONI ET AL. | 3

mental health symptoms, although non‐Veterans or those with no demonstrated strong internal consistency in community (α = .91,
history of service (N = 26) and Veterans without mental health Scher et al., 2001) and clinical (α = .95, Bernstein et al., 1994)
symptoms (N = 9) were also allowed to participate as comparison adult samples, as well as high test‐retest reliability (r = 0.88,
groups and to enhance the generalizability of study findings. This Bernstein et al., 1994).
study used all available data collected at the time of analysis. Note
that a limited dataset from this group (N = 36 at the intake
assessment) was merged with other data and reported in a study 2.2.2 | Depression, anxiety and PTSD symptoms
that examined the validity of the QUIC across multiple age groups
and populations (Glynn et al., 2019). Data utilized for the current Depression and anxiety symptoms in the past 2 weeks were assessed
analysis were from the initial baseline visit (N = 156, and n = 147 for using the nine‐item Patient Health Questionnaire‐9 (PHQ‐9; Kroenke
all participants with both QUIC and CTQ measures), in addition to et al., 2001) and the seven‐item Generalized Anxiety Disorder Scale‐
secondary analyses examining associations of early‐life 7 (GAD‐7; Spitzer et al., 2006), respectively. Both the PHQ‐9 and
unpredictability and symptom change over time, which utilized data GAD‐7 have demonstrated strong internal consistency and
from up to three visits (Ns = 156, 100, 67, at visits 1, 2, 3). Race and test–retest reliability in medical and psychiatric patient samples
ethnicity of the sample population were 8.3% American–Indian/ (Beard & Björgvinsson, 2014; Beard et al., 2016; Kroenke et al., 2001;
Alaskan Native, 8.3% Asian, 21.8% Black, 2.6% Native Hawaiian/ Spitzer et al., 2006). PHQ‐9 scores of 0–4 indicate no symptoms, 5–9
Pacific Islander, 3.2% other/unknown, 65.4% White, and of the total mild symptoms, 10–14 moderate symptoms, 15–19 moderately‐
sample 21.2% were Hispanic/Latino (Table 1a). The CESAMH PTSD/ severe symptoms, and 20–27 severe depression symptoms. GAD‐7
TBI Biorepository Study was approved by the Veterans Affairs cutoff scores of 5, 10, and 15 also indicate mild, moderate, and
Institutional Review Board and all participants provided written, severe anxiety symptoms. PTSD symptoms in the past month were
informed consent. assessed using the PTSD checklist for DSM‐5 (PCL‐5; Weathers et al.
2012). The PCL‐5 has demonstrated strong internal consistency
(α = .96) and test–retest reliability among Veterans (r = 0.84; Bovin
2.2 | Measures et al., 2016). PCL‐5 symptoms >33 suggest the likelihood of PTSD.

2.2.1 | Early‐life adversity

2.2.3 | Anhedonia
Early‐life unpredictability was assessed using the QUIC (Glynn et al.,
2019; https://contecenter.uci.edu/questionnaire-of-unpredictability- To assess anhedonia, we used the Anhedonic Depression subscale of
in-childhood/). The QUIC is a self‐report measure that asks the Mood and Anxiety Symptoms Questionnaire (MASQ‐AD; Watson
respondents about experiences related to caregiving and the & Clark, 1991). This 22‐item scale measures low positive emotion and
household environment before age 18, with a subset of questions anhedonia in the past week and can be used independently of the full
focusing on experiences more likely to occur in earlier childhood MASQ. The MASQ‐AD has demonstrated strong internal consistency
(before age 12). Example items include “I experienced changes in my (α = .90; Kendall et al., 2016).
custody arrangement,” “At least one of my parents regularly checked
that I did my homework,” and “At least one of my parents was
disorganized.” The scale is comprised of 38 items which are endorsed 2.2.4 | Suicidal ideation
as either “yes” or “no” (some of which are reverse coded) and are then
fit to five subscales: parental involvement (9 items), parental To specifically assess QUIC relationships with suicidal ideation (SI),
predictability (12 items), parental environment (7 items), physical we used one item from the PHQ‐9 (“thinking that you would be
environment (7 items), and safety and security (3 items). Total scores better off dead or that you want to hurt yourself in some way”) and
range from 0 to 38, with a higher score indicating greater exposure to one item from the MASQ‐AD (“thought about death or suicide”). Each
unpredictability in the environment before 18 years of age. The QUIC item was examined independently.
total score has demonstrated strong internal consistency (α = .90–.92
among adults) and excellent test–retest reliability (r = 0.92; Glynn
et al., 2019). 2.2.5 | Substance use
Participants also completed the CTQ (Bernstein et al., 1994),
a 28‐item scale that assesses child abuse and neglect (possible Alcohol abuse was assessed using the Alcohol Use Disorders
score from 5 to 125) and includes five subscales: physical, Identification Test (AUDIT; Saunders et al., 1993). This scale allows
emotional and sexual abuse, and physical and emotional neglect. for the assessment of both consumption and dependence behaviors.
A higher score on the CTQ denotes greater endorsement of abuse AUDIT scores >8 but <15 indicate hazardous drinking, while scores
and neglect before the age of 18. The CTQ total score has >14 indicate the likelihood of dependence.
4 | SPADONI ET AL.

TABLE 1 Descriptive statistics for symptoms, trauma exposure, and population demographics

(a) Descriptive statistics for race/ethnicity, early‐life adversity and symptom endorsement in VA TBI/PTSD Biorepository Sample

Number of subjects 156 100 67

Assessment visit 1 2 3

Race/ethnicity

American–Indian/Alaskan Native 8.33% 11.00% 10.45%

Asian 8.33% 8.00% 4.48%

Black 21.79% 24.00% 23.88%

Native Hawaiian/Pacific Islander 2.56% 2.00% 5.97%

White 65.38% 65.00% 64.18%

Other/unknown 3.21% 6.00% 2.99%

Hispanic/Latino 21.15% 19.00% 19.40%

Veteran 77.56% 74.00% 65.67%

Life exposure checklist “happened to me” 8.74 (5.74) 9.24 (5.81) 9.43 (5.63)

Combat Exposure Scale (DRRI‐CES) 27.75 (16.46) a

– –

Post Battle Experiences Scale (DRRI‐PBE) 23.58 (12.97) b

– –

Early‐life unpredictability (QUIC) 12.87 (9.13) – –

Childhood trauma (CTQ) 51.80 (20.20) – –

Anhedonia (MASQ‐22) 70.28 (16.82) 67.80 (18.10) 72.08 (16.45)

PTSD (PCL‐5) 35.17 (21.03) 31.35 (20.72) 33.91 (22.39)

Depression (PHQ‐9) 10.96 (6.97) 9.23 (6.55) 10.38 (7.12)

Anxiety (GAD‐7) 9.14 (6.43) 7.92 (6.07) 8.62 (6.38)

Alcohol use (AUDIT) 3.39 (4.83) 3.46 (5.19) 4.87 (6.94)

Pain intensity (PROMIS) 7.87 (2.64) 7.61 (2.72) 7.65 (2.81)

Pain interference (PROMIS) 16.77 (6.94) 15.16 (6.92) 15.84 (7.40)

Suicidal ideation item (MASQ) 1.53 (0.93) 1.55 (0.98) 1.65 (0.98)

Suicidal ideation item (PHQ‐9) 0.37 (0.68) 0.33 (0.71) 0.35 (0.72)

(b) Education, employment, and income characteristics at each study visit

Number of subjects 156 100 67

Assessment visit 1 2 3

Education

Some high school 1.92% 2.00% 2.99%

GED 1.92% 5.00% 5.97%

High school diploma 7.69% 7.00% 8.96%

Some college 32.69% 32.00% 26.87%

Associates degree 16.03% 13.00% 14.93%

4‐year college degree 25.00% 28.00% 26.87%

Master's degree 12.82% 10.00% 10.45%

Doctoral degree 1.28% 2.00% 2.99%

Not reported 0.64% 1.00% 0.00%

Employment status

Full time 26.28% 25.00% 20.90%

Part time 12.82% 19.00% 14.93%

SPADONI ET AL. | 5

TABLE 1 (Continued)

(b) Education, employment, and income characteristics at each study visit

Number of subjects 156 100 67
Assessment visit 1 2 3

Seasonally 3.85% 1.00% 4.48%

Day labor 0.64% 1.00% 2.99%

Unemployed 55.77% 53.00% 56.72%

Unknown 0.64% 1.00% 0.00%

Income

<$15,000 22.44% 26.00% 23.88%

$15,000–$29,999 17.31% 13.00% 20.90%

$30,000–$44,999 16.03% 22.00% 13.43%

$45,000–$59,999 14.10% 15.00% 17.91%

$60,000–$74,999 7.69% 7.00% 4.48%

$75,000–$99,999 10.90% 9.00% 11.94%

$100,000+ 10.26% 6.00% 4.48%

Note: (a) Mean (standard deviation) per visit of self‐reported measures of anhedonia, depression, anxiety, posttraumatic stress, alcohol use, pain, and
suicidal ideation, as well as prior trauma exposures (LEC, DRRI CES, and DRRI PBE). Assessment Visits 2 and 3 were approximately 3 and 6 months after
the initial assessment visit. See Section 2 for instrument ranges and cutoff scores indicating symptom severity.
(b) Data presented as a percentage of the total population at the specific visit reported.
Abbreviations: AUDIT, Alcohol Use Disorders Identification Test; CTQ, Childhood Trauma Questionnaire; DRRI, Deployment Risk and Resiliency
Inventory; GAD‐7, Generalized Anxiety Disorder Scale‐7; GED, general educational development; LEC, Life Events Checklist for DSM‐5; MASQ‐22, Mood
and Anxiety Symptom Questionnaire‐22; PCL‐5, PTSD Checklist for DSM‐5; PHQ‐9, Patient Health Questionnaire‐9; PROMIS, Patient‐Reported
Outcomes Measurement Information System; PTSD, posttraumatic stress disorder; QUIC, Questionnaire of Unpredictability in Childhood; TBI,
traumatic brain injury; VA, Veterans Affairs.
a
n = 130.
b
n = 129.

2.2.6 | Pain symptoms separate from CTQ contribution (n = 147 subjects

completed both measures). We constructed separate linear
To assess associations with pain symptoms, we used the nine‐item regression models (lm function in R) with a fixed main effect of
Patient‐Reported Outcomes Measurement Information System for QUIC total, and nine dependent variables: total scores on the
pain (PROMIS; www.NIHPROMIS.org). The PROMIS measures were PCL‐5 (PTSD symptoms), GAD‐7 (anxiety symptoms), PHQ‐9
developed using item response theory and calibrated in a sample of (depression symptoms), MASQ‐AD (anhedonia), AUDIT (alcohol
21,133 people, with the aim of providing highly reliable, precise use), and PROMIS current pain (interference and intensity
measures of patient‐reported health status for physical, mental, and scores). In the case of MASQ‐AD Item 22 and PHQ‐9 Item 9
social well‐being. Here, we focused on the pain intensity (range of (SI), we applied a Poisson distribution (glm function in R), as the
3–15, a score of 8 is approximately equivalent to average pain variables were highly positively skewed. A Benjamini–Hochberg
intensity endorsed by the US population) and interference (possible correction for multiple comparisons was applied for the single‐
range 6–30, a score of 8 is approximately equivalent to average pain level regressions (nine models). We repeated these analyses on
interference endorsed by US population) subscales. whitened QUIC totals in regressions that were found to be
significant (5 total). Finally, hierarchical regressions were also
used to examine the specific contribution of the QUIC to R2 after
2.3 | Analyses accounting for the CTQ in measure variance.
To examine the effects of early‐life unpredictability, time, and
All statistical analyses were performed using R Studio, version their interaction on measures of symptom severity, we con-
1.2.5042. In Time 1 analyses which compare QUIC and CTQ total structed separate linear mixed effect models with a random
scores as predictors of symptom severity, due to the high intercept to account for within‐subject correlation on the
covariance of QUIC and CTQ scores, we utilized a whitening measures (lmer function in R). As in the single‐level regressions,
technique (Kessy et al., 2018) to quantify QUIC associations with a Poisson distribution was used to model the SI item. For
6 | SPADONI ET AL.

examination of the associations of QUIC with symptoms over TABLE 2 Treatment seeking characteristics
time, we utilized participants in the sample with up to three visits. Overall percent of participants receiving current mental
Fixed main effects included mean‐centered QUIC total, visit health treatment 65.38%
number (i.e., −1, 0, 1), and a fixed interaction for mean‐centered PTSD 72.55%
QUIC total and visit number. Time was coded as nonoverlapping
Depression 64.71%
ordinal levels (1 = baseline [n = 156]; 2 = 3.23 months
[range = 1.40–4.70 months; n = 100]; 3 = 7.08 months Anxiety 55.88%

[range = 4.80–10.27 months; n = 67]). Five predictor–outcome Schizophrenia/psychosis 3.92%

relationships were modeled independently (i.e, PCL‐5 [PTSD Bipolar disorder 7.84%
symptoms], GAD‐7 [anxiety symptoms], PHQ‐9 [depression
Substance/alcohol abuse 4.90%
symptoms], MASQ‐AD [anhedonia], as well as PHQ‐9 Item 9
[SI]). The random effect variance–covariance matrix was Other 5.88%

unstructured with a random intercept. A Benjamini–Hochberg Note: The majority of Veterans (n = 102/156) self‐identified as currently
correction for multiple comparisons was applied for the associa- receiving treatment at Visit 1. Within those individuals, the most common
reasons for seeking treatment were PTSD, depression, and anxiety.
tions of the QUIC with symptoms over time (five models).
Abbreviation: PTSD, posttraumatic stress disorder.

3 | RESULTS
unpredictability and pain symptoms (PROMIS pain intensity and
3.1 | Sample demographics interference) (Fs[1, 144] ≤ 1.06, p's > 0.31). We also tested for
interactions between unpredictability and gender on these same
The participant population was majority male (62%), white (65%), and outcome variables. There was a trend level interaction for an
currently seeking treatment for PTSD and/or depression (65%). unpredictability by gender interaction on the PCL‐5 (p = .054). When
Participants ranged in age from 21 to 90 years old. They endorsed correlations were completed within each gender, only women
moderate levels of childhood unpredictability (Glynn et al., 2019) and showed a significant association between QUIC scores and PTSD
were above clinical cutoffs for moderate childhood trauma (Bernstein symptom severity as measured by PCL‐5 (Supporting Information:
& Fink, 1998; see Table 1). The majority reported they were in active Figure S1; women: r = 0.40, p = .001; men: r = 0.08, p = .45). There
treatment (behavioral and/or pharmacotherapy) for PTSD, depres- were no significant gender differences in PCL‐5 total scores (F
sion, and/or substance use (Table 2, and see Supporting Information: (1,145) = 2.4, p = .12), although we did observe that women (n = 59)
Table S1 for a more detailed description of treatment types across had slightly lower PCL‐5 total scores than men (n = 88), (i.e., mean
the sample population). (SD)women = 31.95 (22.70), mean (SD)men = 37.45 (20.00) (see box-
plots in Supporting Information: Figure S2). Komolgorov–Smirnov
tests also indicated no difference between the sex‐specific distribu-
3.2 | Association of early‐life unpredictability with tions (D = 0.180, p = .204). Furthermore, after controlling for the
psychiatric symptoms and pain influence of CTQ total scores, the gender by QUIC interaction was
significant (p = .04), and there was no interaction between CTQ and
The results of the linear regressions indicated that early‐life gender on PCL scores (p = .57). There were no significant interactions
unpredictability (QUIC total) significantly predicted greater baseline between gender and anxiety (GAD‐7), depression (PHQ‐9), or
PTSD symptoms (PCL‐5) (β = .214, t = 2.64, p = .009, adjusted [Adj.] anhedonia (MASQ‐AD). See Table 3 for all models.
R2 = 0.039, F(1,145) = 6.94, p = .009), anxiety (GAD‐7) (β = .250,
t = 3.10, p = .002, Adj. R2 = 0.056, F(1,145) = 9.63, p = .002), depres-
sion (PHQ‐9) (β = .247, t = 3.07, p = .003, Adj. R2 = 0.055, F 3.3 | Association of early‐life unpredictability
(1,145) = 9.43, p = .003), and anhedonia (MASQ‐AD) (β = .258, with SI
t = 3.21, p = .002, Adj. R2 = 0.060, F(1,145) = 10.32, p = .002). See
Figure 1 for correlation plots. We did not observe significant variance To test the specific contribution of early‐life unpredictability to
accounted for by early‐life unpredictability (QUIC) on alcohol use current SI, we performed two Poisson regressions given that both
(AUDIT scores) (F[1, 143] = 0.058, p = .809). Upon examination of the measures of SI were highly positively skewed. The first regression
data, we did find a relatively high number of participants who examined the association between QUIC total and PHQ‐9 Item 9
endorsed no recent drinking (n = 64; 41%) likely due to participants’ (“Thoughts that you would be better off dead, or thoughts of hurting
enrollment in PTSD and/or alcohol treatment programs. However, yourself in some way?”). The percent change in SI was 4.4% (95%
we also did not detect a relation between alcohol use and confidence interval [CI] = 1.7–7.0) higher for every unit increase in
unpredictability in the subset of participants who endorsed drinking QUIC total (z = 3.35, p = .0008). We used the residual deviance to
(ρ = −0.002, p = .98). We also did not detect a relation between perform a goodness of fit test for the overall model. The goodness‐
SPADONI ET AL. | 7

F I G U R E 1 Early‐life unpredictability is significantly positively associated with symptoms of anhedonia, depression, and anxiety. Scatterplots
of and regression results of QUIC total score associations with PCL‐5, GAD‐7, PHQ‐9, and MASQ‐AD total scores:. Trauma‐related symptoms
(PCL‐5 total scores) β = .214, t = 2.64, p = .009, Adj. R2 = 0.039, F(1,145) = 6.94, p = .009. Anxiety (GAD‐7 total scores) β = .250, t = 3.10, p = .002,
Adj. R2 = 0.056, F(1,145) = 9.63, p = .002. Depression (PHQ‐9 total scores) β = .247, t = 3.07, p = .003, Adj. R2 = 0.055, F(1,145) = 9.43, p = .003.
Anhedonic depression (MASQ‐AD total scores) β = .258, t = 3.21, p = .002, Adj. R2 = 0.060, F(1,145) = 10.32, p = .002. Adj., adjusted; GAD‐7,
Generalized Anxiety Disorder Scale‐7; MASQ‐AD, Mood and Anxiety Symptom Questionnaire‐Anhedonic Depression; PCL‐5, PTSD Checklist
for DSM‐5; PHQ‐9, Patient Health Questionnaire‐9; PTSD, posttraumatic stress disorder; QUIC, Questionnaire of Unpredictability in Childhood.

of‐fit χ2 test was not statistically significant (res. dev = 142.3719, variables into orthogonal variables (i.e., r‐values between symptom
p = .55) indicating good model fit (https://stats.idre.ucla.edu/r/dae/ clusters will be set to zero). This whitening approach also maximizes
poisson‐regression/). A second Poisson regression was run to predict the correlation between the original variables and the newly
SI as measured by MASQ‐AD Item 22 (“Thought about death or transformed “whitened” variables (i.e., new r > .78). Therefore, the
suicide”) from QUIC total. The contribution of the MASQ‐AD Item 22 new whitened QUIC and CTQ totals continue to measure the same
to the overall model did not meet the conventional threshold for constructs, but they are no longer correlated with each other. This
statistical significance (z = 1.68, p = .09). Finally, we applied a approach mitigates the issue of multicollinearity and allows us to test
Benjamini–Hochberg correction for the nine models testing the for specific incremental validity of the QUIC in understanding how
association between QUIC and depression, anxiety, trauma symp- early‐life unpredictability contributes to adult psychopathology.
toms, anhedonia, alcohol use, pain severity (two items), and suicidality We reran regressions using both the whitened QUIC and CTQ
(two items), and significant results were retained as reported total scores as simultaneous predictors of those factors that we
(ps < .045). See Table 3a for regression coefficients and significance found to be significantly predicted by the QUIC alone: PTSD
values. symptoms (PCL‐5), anxiety symptoms (GAD‐7), depression symp-
toms (PHQ‐9), anhedonia (MASQ‐AD), and SI (PHQ‐9 Item 9). The
whitened QUIC total significantly predicted anxiety (GAD‐7)
3.4 | Early‐life unpredictability compared to (QUIC β = .176, t = 2.18, p = .031; CTQ β = .185, t = 2.30, p = .023;
childhood trauma overall model Adj. R2 = 0.052, F(2,144) = 5.02, p = .008), depres-
sion (PHQ‐9) (QUIC β = .159, t = 1.976, p = .05; CTQ β = .206,
Our ability to tease apart the specific contribution of early‐life t = 2.55, p = .01; overall model Adj. R2 = 0.055, F(2,144) = 5.21,
unpredictability to the development of psychopathology may be p = .007), and anhedonia (MASQ‐AD) (QUIC β = .192, t = 2.39,
limited by its high correlation with other measures of early adversity, p = .018; CTQ β = .176, t = 2.18, p = .03; overall model Adj.
including abuse and neglect. For example, we observe a correlation of R2 = 0.055, F(2,144) = 5.24, p = .006). Interestingly, for the PCL‐
r = .75 between QUIC and CTQ total scores. To address this issue, we 5, the QUIC was no longer a significant contributor to the overall
performed a second set of regressions after conducting a statistical model once CTQ was added, while CTQ was a significant
whitening procedure with the R package: whitening (Kessy et al., predictor (QUIC β = .128, t = 1.57, p = .118; CTQ β = .192,
2018). This approach transformed the original QUIC and CTQ t = 2.37, p = .019; overall model Adj. R2 = 0.040, F(2,144) = 4.05,
8 | SPADONI ET AL.

TABLE 3 Statistical results for models examining early‐life unpredictability associations with psychiatric symptoms

(a) Early‐life unpredictability alone

QUIC Overall regression

β t p Adjusted R2 F p

Anxiety (GAD‐7) .25 3.10 .002 0.056 9.63 .002

Depression (PHQ‐9) .247 3.07 .003 0.055 9.43 .003

Anhedonia (MASQ‐AD) .258 3.21 .002 0.06 10.3 .002

PTSD (PCL‐5) .214 2.64 .009 0.039 6.94 .009

%ΔSuicidal ideation (SI) z p

SI (PHQ‐9) 4.4% (1.7–7.0) 3.35 .0008

SI (MASQ) 1.68 .094

(b) Early‐life unpredictability compared to childhood trauma

QUIC CTQ Overall regression
β t p β t p Adjusted R2 F p ΔR2b

Anxiety (GAD‐7) .176 2.18 .031 .185 2.30 .023 0.052 5.02 .008 0.031

Depression (PHQ‐9) .159 1.976 .05 .206 2.55 .01 0.055 5.21 .007 0.025

Anhedonia (MASQ‐AD) .192 2.39 .018 .176 2.18 .03 0.055 5.24 .006 0.037

PTSD (PCL‐5) .128 1.57 .118 .192 2.37 .019 0.04 4.05 .019 NS

%ΔSIa z p %ΔSI z p

SI (PHQ‐9) 2.15 (2.74–4.58) 1.74 0.082 4.0 (1.32–6.68) 3.15 .002 NS

(c) Early‐life unpredictability associations with symptoms when collapsed across three visits
QUIC
βa,c t p

Anxiety (GAD‐7) .16 3.21 .002

Depression symptoms .18 3.13 .002

(PHQ‐9)

Anhedonia (MASQ‐AD) .53 3.8 .0002

PTSD symptoms (PCL‐5) .51 2.82 .005

SI (PHQ‐9 Item 9) .04 2.17 .03

Note: Table depicts the original statistical model to examine the contribution of early life unpredictability to psychiatric symptoms at Visit 1 (a) followed by
models comparing the contribution of unpredictability and childhood trauma at Visit 1 (b) and the main effects of unpredictability across all three visits (c).
a
%ΔSI = percent change in SI for every unit increase in QUIC total.
b
ΔR2 = We conducted a series of hierarchical regressions in R Studio to inspect the change in R‐square (ΔR2) with the addition of the QUIC to the CTQ in
explaining variance in our primary outcome variables. See Section 3, for details.
Abbreviations: CTQ, Childhood Trauma Questionnaire; GAD‐7, Generalized Anxiety Disorder Scale‐7; MASQ‐AD, Mood and Anxiety Symptom
Questionnaire‐Anhedonic Depression; NS, not significant; PCL‐5, PTSD Checklist for DSM‐5; PHQ‐9, Patient Health Questionnaire‐9; PTSD,
posttraumatic stress disorder; QUIC, Questionnaire of Unpredictability in Childhood.
c
The lmer function does not provide a standardized β value.

p = .019). The contribution of early‐life unpredictability (QUIC Studio to inspect the change in R square (Δ R2 ) with the addition
total score) to the overall model predicting PHQ‐9 item 9 (SI) was of the QUIC to the CTQ in explaining variance in our primary
diminished when considering the contribution of the CTQ. For outcome variables. Consistent with the simultaneous regressions,
the QUIC, the percent change in reported SI was 2.15% (95% we found that the addition of the QUIC significantly improved
CI = −2.74 to 4.58) higher for every unit increase (z = 1.74, model fit for depression (PHQ‐9 total; ΔR2 = 0.025), anxiety
p = .082). For the CTQ, the percent change in reported SI was (GAD7; ΔR 2 = 0.031), and anhedonia (MASQAD; ΔR2 = 0.037).
4.0% (95% CI = 1.32–6.68) higher for every unit increase (z = 3.15, Scores on the CTQ alone did not predict alcohol (AUDIT) or
p = .002). We conducted a series of hierarchical regressions in R current pain (PROMIS pain intensity or interference), consistent
SPADONI ET AL. | 9

with the QUIC findings. See Table 3b for regression coefficients unpredictability and self‐reported measures of PTSD symptoms,
and significance values. anxiety symptoms, depression symptoms, anhedonia, and SI. Taken
together, these data suggest that early‐life unpredictability is
uniquely associated with anxiety and mood‐related symptoms among
3.5 | Early‐life unpredictability associations with adults with trauma exposure. While early‐life unpredictability was
symptoms across time also related to trauma‐related symptoms and SI, childhood trauma
was a stronger predictor, supporting a specific relationship between
Finally, we explored whether the QUIC predicted symptom states in childhood traumatic experiences and adult trauma symptoms as well
those measures that showed a significant relationship with the QUIC as SI (Angelakis et al., 2019; Brewin et al., 2000). Importantly, the
at Time 1 (i.e., PTSD, anxiety, depression, anhedonia, and suicidality data support accumulating evidence for conceptualizing adversity
[PHQ‐9 SI]) over a 6‐month period. Results of the linear mixed model across distinct dimensions including unpredictability.
with QUIC total and time as fixed effects indicated that QUIC total The current work provides additional evidence that early‐life
significantly predicted PTSD symptoms (β = .51, t = 2.82, p = .005), unpredictability may be an important type of early‐life adversity that
anxiety symptoms (β = .16, t = 3.21, p = .002), depression symptoms influences the risk for psychopathology. This study extends existing
(β = .18, t = 3.13, p = .002), and anhedonia (β = .53, t = 3.80, p = .0002) models of how childhood adversity may be linked to psycho-
across all three time points. In each case, the main effects signaled pathology (e.g., cumulative stress; Evans et al., 2013) by providing
that higher QUIC total scores were related to consistently elevated evidence that unpredictability is a rarely considered but impactful
scores on all outcome measures (and exceeded Bonferroni correc- form of negative early‐life experience. In other words, early‐life
tions of p ≤ .0125). There was also a main effect of time on PCL‐5 unpredictability may help explain the powerful and pervasive
scores (β = −1.71, t = −2.30, p = .02), such that PCL‐5 total scores association between adversity and psychopathology by highlighting
decreased overall across time, which would be expected given that that it is not only the quality of early life experiences but also, the
the majority of patients were receiving active treatment (65%) (See underlying patterns of caregiving and environmental signals that are
Table 2 for treatment‐seeking characteristics, and Table S1 for key (for review, see Glynn & Baram, 2019). As such, the addition of
treatment modality characteristics). There were no other significant unpredictability to our existing models can enrich our understanding
main effects of time, or interactions between QUIC and time, on our of how childhood adversity leads to psychopathology, and bolster our
variables of interest. For SI, results of the generalized linear mixed ability to predict who is at greatest risk for psychiatric outcomes and
models (R function “glmer,” family= Poisson) with QUIC total and time which outcomes may be most likely (e.g., Kessler et al., 2010), as well
as fixed effects indicated that QUIC total significantly predicted SI on as guide the development of targeted interventions (McLaughlin
PHQ‐9 Item 9 (β = .04, t = 2.17, p = .03). There were no main effects et al., 2019).
of time, or significant interactions of QUIC with time, for either SI In the current sample of adults with trauma exposure, we found
outcome. These results survived correction for multiple comparisons that endorsement of higher levels of early‐life unpredictability was
(five variables) after applying a Benjamini–Hochberg adjustment associated with increased anhedonia, above and beyond childhood
(ps ≤ .03). See Table 3c for regression coefficients and significance trauma. These data support the increasing evidence that
values. unpredictability (as measured here by household routine and
caregiving predictability) may be linked to disruption in reward
processes (Birnie et al., 2020). Rodent models of fragmented
4 | DISC US SION maternal care consistently result in relatively specific anhedonia‐
like phenotypes, such as reduced interest in social and appetitive
The principal findings of the current study testing associations reward stimuli (Bolton et al., 2018; Molet, Heins, et al., 2016), that are
between early‐life unpredictability and psychiatric symptoms in a causally linked to aberrant signaling in the amygdala‐prefrontal cortex
sample of adults with trauma exposure are: (1) greater early‐life and amygdala‐nucleus accumbens circuits (Birnie et al., 2020; Bolton
unpredictability is associated with greater PTSD symptoms, anxiety et al., 2018). It will be critical for future studies to examine in what
symptoms, depression symptoms, and anhedonia, and increased risk manner anhedonia may be most associated with early‐life
for SI at the baseline assessment. (2) These associations are specific, unpredictability to better understand the mechanisms that may
as the QUIC does not significantly associate with alcohol use or pain underlie the relationship between this form of adversity and
symptoms. (3) The use of a whitening approach to reduce multi- psychiatric symptoms.
collinearity between measures of early‐life unpredictability and There are a number of potential mechanisms through which
childhood trauma, in conjunction with hierarchical regressions, early‐life unpredictability may modify stress response systems and
revealed that early‐life unpredictability significantly contributes to circuits. Exposure to higher unpredictability of maternal sensory
the variance of anxiety symptoms, depression symptoms, and signals in infancy is associated with blunted infant cortisol to a painful
anhedonia above and beyond the contribution of childhood trauma. stressor at one year old (Noroña‐Zhou et al., 2020), suggesting that
(4) The described associations persisted for a minimum of 6 months, this unpredictability can shape hypothalamic–pituitary–adrenal (HPA)
indicating a consistent positive association between early‐life axis responding in early development. In animals, fragmented early
10 | SPADONI ET AL.

care or stress during early life consistently results in long‐term unpredictability, it should be noted this instrument has been validated
disruption of the HPA axis and its primary signaling system using prospective observational data of early‐life unpredictability as
corticotrophin‐releasing factor, resulting in abnormal neuronal well as prospective validation of items on the QUIC (e.g., moved
function across multiple corticolimbic circuits in adulthood (Chen & frequently; Glynn et al., 2019), supporting its reliability in measuring
Baram, 2016; Gunn et al., 2013; Toth et al., 2016). In animals and this newly conceptualized dimension of early‐life adversity. (3) While
humans, unpredictable sensory signals from caregivers alter the there was significant participant attrition over time, sample attributes
hippocampal structure, memory function, and executive function remained steady across visits (see Table 1), suggesting overall stability
(Davis et al., 2019, 2017), as well as connectivity to cortical circuits of participant characteristics. (4) Finally, null associations between
(Granger et al., 2021). These structural changes are potentially linked QUIC and alcohol use and pain should be interpreted cautiously given
to altered synapse number and function in the hippocampus, the relatively low base rates of these symptoms in this sample (only
observed in animal models of unpredictable care (Ivy et al., 2008; about half the sample currently consumed alcohol); pain intensity
Molet, Maras, et al., 2016). Reduced hippocampal function and scores were similar to the average score for the US population
abnormalities in structure and connectivity are a consistent pheno- indicating a potentially restricted range within which to detect
type for both depression and PTSD, with some indication that it may associations.
be a predisposing risk factor for these disorders (for review, see
Acheson et al., 2012). An additional corticolimbic circuit that may be
disrupted by unpredictable care is the uncinate fasciculus, the 5 | CONCLUSION
primary fiber bundle connecting the amygdala to the orbitofrontal
cortex and a key component of the medial temporal lobe prefrontal These findings indicate that early‐life unpredictability is associated
cortex circuit. Exposures to higher levels of unpredictable maternal with mood and anxiety symptoms, as well as comorbid SI and
signals in infancy predicted greater generalized fractional anisotropy anhedonia, in individuals with clinical levels of mood and anxiety
of the uncinate fasciculus in middle childhood, which in turn was symptoms. These relations are stable over time and are selectively
associated with reduced episodic memory (Granger et al., 2021). above and beyond contributions of other forms of adversity such as
Overall, these studies indicate that unpredictable care may disrupt childhood trauma. These findings support the examination of the
the maturation of critical corticolimbic circuits mediating reward and unique contribution of unpredictability to the development and
memory functions, which may in turn confer risk for psychiatric maintenance of these symptoms across the lifespan.
symptomatology. Notably, the practice of family routines during the
Covid‐19 pandemic predicts lower levels of psychopathology among ACKNOWLEDGME NT S
children (Glynn et al., 2021), suggesting that routine may buffer This study was supported by NIMH P50MH096889 (Andrea D.
against the deleterious effects of unpredictability, though the precise Spadoni, Laura M. Glynn, Elysia P. Davis, Tallie Z. Baram, Dewleen G.
mechanisms remain unclear. Aberrant social information and emotion Baker, and Victoria B. Risbrough), VA Merit Awards (Andrea D.
processing, learning, and accelerated biological aging are proposed as Spadoni 5I01CX001762 and Victoria B. Risbrough 5I01BX004312),
other possible transdiagnostic mechanisms underlying associations the Office of Academic Affiliations, Advanced Fellowship Program in
between early adversity and psychopathology (McLaughlin Mental Illness Research and Treatment, Department of Veterans
et al., 2020, 2019). Further examination of the psychological and Affairs (Meghan Vinograd), and the Center of Excellence for Stress
biological mechanisms underlying the association between early‐life and Mental Health (Andrea D. Spadoni, Meghan Vinograd, Dewleen
unpredictability and psychiatric outcomes is an important direction G. Baker, Caroline M. Nievergelt, and Victoria B. Risbrough).
for future research because these processes may be effective targets
of intervention. CONFLIC T OF INTEREST
This study has a number of strengths and limitations. The Victoria B. Risbrough has received consulting fees from Engrail, Jazz
strengths are in the relatively large sample size as compared to earlier Pharmaceuticals, and Fallon Capital in the last 36 months.
inquiries of the measure, associations with a longitudinal assessment of
psychiatric symptoms, and the wide range of early‐life unpredictability DATA AVAILABILITY STATEMENT
and symptom severity in the study population. These strengths The data that support the findings of this study are available from the
support the generalizability of the findings. Limitations are that (1) corresponding author upon reasonable request.
this is a majority Veteran (77%), White (65%), and male (62%)
population, (2) the study relied on a retrospective measurement of ORC I D
early‐life unpredictability, (3) there was significant attrition over study Victoria B. Risbrough http://orcid.org/0000-0001-8347-3820
visits, (4) we did not find associations between QUIC and alcohol‐
related problems and pain ratings. Therefore, (1) additional work with a RE F ER EN CES
more diverse range of individuals is necessary to assess how the QUIC Acheson, D. T., Gresack, J. E., & Risbrough, V. B. (2012). Hippocampal
performs depending on a variety of personal characteristics. (2) While dysfunction effects on context memory: Possible etiology for
posttraumatic stress disorder. Neuropharmacology, 62(2), 674–685.
the study relied on a retrospective measurement of early‐life
SPADONI ET AL. | 11

Agorastos, A., Pittman, J. O. E., Angkaw, A. C., Nievergelt, C. M., Hansen, C. J., Psychopathology, 28(4 Pt 2), 1505–1516. https://doi.org/10.1017/
Aversa, L. H., Parisi, S. A., Barkauskas, D. A., & Baker, D. G. (2014). The S0954579415001169
cumulative effect of different childhood trauma types on self‐reported Evans, G. W., Li, D., & Whipple, S. S. (2013). Cumulative risk and child
symptoms of adult Male depression and PTSD, substance abuse and development. Psychological Bulletin, 139(6), 1342–1396. https://doi.
health‐related quality of life in a large active‐duty military cohort. Journal org/10.1037/a0031808
of Psychiatric Research, 58, 46–54. Frewen, P. A., Dozois, D. J. A., & Lanius, R. A. (2012). Assessment of
Angelakis, I., Gillespie, E. L., & Panagioti, M. (2019). Childhood anhedonia in psychological trauma: Psychometric and neuroimaging
maltreatment and adult suicidality: A comprehensive systematic perspectives. European Journal of Psychotraumatology, 3, 8587.
review with meta‐analysis. Psychological Medicine, 49(7), https://doi.org/10.3402/ejpt.v3i0.8587
1057–1078. Glynn, L. M., & Baram, T. Z. (2019). The influence of unpredictable,
Beard, C., & Björgvinsson, T. (2014). Beyond generalized anxiety disorder: fragmented parental signals on the developing brain. Frontiers in
Psychometric properties of the GAD‐7 in a heterogeneous Neuroendocrinology, 53, 100736. https://doi.org/10.1016/j.yfrne.
psychiatric sample. Journal of Anxiety Disorders, 28(6), 547–552. 2019.01.002
https://doi.org/10.1016/j.janxdis.2014.06.002 Glynn, L. M., Davis, E. P., Luby, J. L., Baram, T. Z., & Sandman, C. A. (2021).
Beard, C., Hsu, K. J., Rifkin, L. S., Busch, A. B., & Björgvinsson, T. (2016). A predictable home environment May protect child mental health
Validation of the PHQ‐9 in a psychiatric sample. Journal of Affective during the COVID‐19 pandemic. Neurobiology of Stress, 14, 100291.
Disorders, 193, 267–273. https://doi.org/10.1016/j.jad.2015.12.075 https://doi.org/10.1016/j.ynstr.2020.100291
Bernstein, D., & Fink. (1998). Childhood trauma questionnaire: A retrospec- Glynn, L. M., Howland, M. A., Sandman, C. A., Davis, E. P., Phelan, M.,
tive self‐report. San Antonio, TX: The Psychological Corporation. Baram, T. Z., & Stern, H. S. (2018). Prenatal maternal mood patterns
Bernstein, D. P., Fink, L., Handelsman, L., Foote, J., Lovejoy, M., predict child temperament and adolescent mental health. Journal of
Wenzel, K., Sapareto, E., & Ruggiero, J. (1994). Initial reliability and Affective Disorders, 228, 83–90. https://doi.org/10.1016/j.jad.2017.
validity of a new retrospective measure of child abuse and neglect. 11.065
American Journal of Psychiatry, 151(8), 1132–1136. https://doi.org/ Glynn, L. M., Stern, H. S., Howland, M. A., Risbrough, V. B., Baker, D. G.,
10.1176/ajp.151.8.1132 Nievergelt, C. M., Baram, T. Z., & Davis, E. P. (2019). Measuring
Birnie, M. T., Kooiker, C. L., Short, A. K., Bolton, J. L., Chen, Y., & novel antecedents of mental illness: The Questionnaire of
Baram, T. Z. (2020). Plasticity of the reward circuitry after early‐life Unpredictability in Childhood. Neuropsychopharmacology, 44(5),
adversity: Mechanisms and significance. Biological Psychiatry, 87(10), 876–882. https://doi.org/10.1038/s41386-018-0280-9
875–884. https://doi.org/10.1016/j.biopsych.2019.12.018 Granger, S. J., Glynn, L. M., Sandman, C. A., Small, S. L., Obenaus, A.,
Bolton, J. L., Molet, J., Ivy, A., & Baram, T. Z. (2017). New insights into Keator, D. B., Baram, T. Z., Stern, H., Yassa, M. A., & Davis, E. P.
early‐life stress and behavioral outcomes. Current Opinion in (2021). Aberrant maturation of the uncinate fasciculus follows
Behavioral Sciences, 14, 133–139. https://doi.org/10.1016/j. exposure to unpredictable patterns of maternal signals. Journal of
cobeha.2016.12.012 Neuroscience, 41(6), 1242–1250. https://doi.org/10.1523/
Bolton, J. L., Molet, J., Regev, L., Chen, Y., Rismanchi, N., Haddad, E., JNEUROSCI.0374-20.2020
Yang, D. Z., Obenaus, A., & Baram, T. Z. (2018). Anhedonia following Gunn, B. G., Cunningham, L., Cooper, M. A., Corteen, N. L., Seifi, M.,
Early‐Life adversity involves aberrant interaction of reward and Swinny, J. D., & Belelli, D. (2013). Dysfunctional astrocytic and
anxiety circuits and is reversed by partial silencing of amygdala synaptic regulation of hypothalamic glutamatergic transmission in a
corticotropin‐releasing hormone gene. Biological Psychiatry, 83(2), mouse model of early‐life adversity: Relevance to neurosteroids and
137–147. https://doi.org/10.1016/j.biopsych.2017.08.023 programming of the stress response. Journal of Neuroscience, 33(50),
Bovin, M. J., Marx, B. P., Weathers, F. W., Gallagher, M. W., Rodriguez, P., 19534–19554. https://doi.org/10.1523/jneurosci.1337-13.2013
Schnurr, P. P., & Keane, T. M. (2016). Psychometric properties of the Howland, M. A., Sandman, C. A., Davis, E. P., Stern, H. S., Phelan, M.,
PTSD checklist for diagnostic and statistical manual of mental Baram, T. Z., & Glynn, L. M. (2020). Prenatal maternal mood entropy
disorders‐fifth edition (PCL‐5) in veterans. Psychological Assessment, is associated with child neurodevelopment. Emotion, 21, 489–498.
28(11), 1379–1391. https://doi.org/10.1037/pas0000254 https://doi.org/10.1037/emo0000726
Brewin, C. R., Andrews, B., & Valentine, J. D. (2000). Meta‐analysis of risk Ivy, A. S., Brunson, K. L., Sandman, C., & Baram, T. Z. (2008).
factors for posttraumatic stress disorder in trauma‐exposed adults. Dysfunctional nurturing behavior in rat dams with limited access
Journal of Consulting and Clinical Psychology, 68(5), 748–766. to nesting material: A clinically relevant model for early‐life
Chen, Y., & Baram, T. Z. (2016). Toward understanding how early‐life stress. Neuroscience, 154(3), 1132–1142. https://doi.org/10.
stress reprograms cognitive and emotional brain networks. 1016/j.neuroscience.2008.04.019
Neuropsychopharmacology, 41(1), 197–206. https://doi.org/10. Kendall, A. D., Zinbarg, R. E., Bobova, L., Mineka, S., Revelle, W.,
1038/npp.2015.181 Prenoveau, J. M., & Craske, M. G. (2016). Measuring positive emotion
Davis, E. P., Korja, R., Karlsson, L., Glynn, L. M., Sandman, C. A., with the mood and anxiety symptom questionnaire: Psychometric
Vegetabile, B., Kataja, E. L., Nolvi, S., Sinervä, E., Pelto, J., properties of the anhedonic depression scale. Assessment, 23(1), 86–95.
Karlsson, H., Stern, H. S., & Baram, T. Z. (2019). Across continents https://doi.org/10.1177/1073191115569528
and demographics, unpredictable maternal signals are associated Kessler, R. C., McLaughlin, K. A., Green, J. G., Gruber, M. J., Sampson, N. A.,
with children's cognitive function. EBioMedicine, 46, 256–263. Zaslavsky, A. M., Aguilar‐Gaxiola, S., Alhamzawi, A. O., Alonso, J.,
https://doi.org/10.1016/j.ebiom.2019.07.025 Angermeyer, M., Benjet, C., Bromet, E., Chatterji, S., de Girolamo, G.,
Davis, E. P., Stout, S. A., Molet, J., Vegetabile, B., Glynn, L. M., Demyttenaere, K., Fayyad, J., Florescu, S., Gal, G., Gureje, O., …
Sandman, C. A., Heins, K., Stern, H., & Baram, T. Z. (2017). Exposure Williams, D. R. (2010). Childhood adversities and adult psychopathology
to unpredictable maternal sensory signals influences cognitive in the WHO World Mental Health Surveys. British Journal of Psychiatry,
development across species. Proceedings of the National Academy 197(5), 378–385. https://doi.org/10.1192/bjp.bp.110.080499
of Sciences of the United States of America, 114(39), 10390–10395. Kessy, A., Lewin, A., & Strimmer, K. (2018). Optimal whitening and
https://doi.org/10.1073/pnas.1703444114 decorrelation. The American Statistician, 72(4), 309–314. https://doi.
Doom, J. R., Vanzomeren‐Dohm, A. A., & Simpson, J. A. (2016). Early org/10.1080/00031305.2016.1277159
unpredictability predicts increased adolescent externalizing behav- Kroenke, K., Spitzer, R. L., & Williams, J. B. (2001). The PHQ‐9: Validity of
iors and substance use: A life history perspective. Development and a brief depression severity measure. Journal of General Internal
12 | SPADONI ET AL.

Medicine, 16(9), 606–613. https://doi.org/10.1046/j.1525-1497. Test (AUDIT): WHO collaborative project on early detection of
2001.016009606.x persons with harmful alcohol consumption–II. Addiction, 88(6),
McGinnis, E. W., Sheridan, M., & Copeland, W. E. (2022). Impact of 791–804.
dimensions of early adversity on adult health and functioning: A 2‐ Scher, C. D., Stein, M. B., Asmundson, G. J., McCreary, D. R., & Forde, D. R.
decade, longitudinal study. Development and Psychopathology, (2001). The Childhood Trauma Questionnaire in a community sample:
https://doi.org/10.1017/S095457942100167X Psychometric properties and normative data. Journal of Traumatic Stress,
McLaughlin, K. A., Colich, N. L., Rodman, A. M., & Weissman, D. G. (2020). 14(4), 843–857. https://doi.org/10.1023/a:1013058625719
Mechanisms linking childhood trauma exposure and psycho- Simpson, J. A., Griskevicius, V., Kuo, S. I., Sung, S., & Collins, W. A. (2012).
pathology: A transdiagnostic model of risk and resilience. BMC Evolution, stress, and sensitive periods: The influence of
Medicine, 18, 96. https://doi.org/10.1186/s12916-020-01561-6 unpredictability in early versus late childhood on sex and risky
McLaughlin, K. A., DeCross, S. N., Jovanovic, T., & Tottenham, N. (2019). behavior. Developmental Psychology, 48(3), 674–686.
Mechanisms linking childhood adversity with psychopathology: Spitzer, R. L., Kroenke, K., Williams, J. B., & Lowe, B. (2006). A brief
Learning as an intervention target. Behaviour Research and Therapy, measure for assessing generalized anxiety disorder: The GAD‐7.
118, 101–109. https://doi.org/10.1016/j.brat.2019.04.008 Archives of Internal Medicine, 166(10), 1092–1097. https://doi.org/
McLaughlin, K. A., & Sheridan, M. A. (2016). Beyond cumulative risk: A 10.1001/archinte.166.10.1092
dimensional approach to childhood adversity. Current Directions in Szepsenwol, O., Simpson, J. A., Griskevicius, V., Zamir, O., Young, E. S.,
Psychological Science, 25(4), 239–245. https://doi.org/10.1177/ Shoshani, A., & Doron, G. (2021). The effects of childhood
0963721416655883 unpredictability and harshness on emotional control and relationship
Molet, J., Heins, K., Zhuo, X., Mei, Y. T., Regev, L., Baram, T. Z., & Stern, H. quality: A life history perspective. Development and Psycho-
(2016). Fragmentation and high entropy of neonatal experience pathology, https://doi.org/10.1017/S0954579421001371
predict adolescent emotional outcome. Translational Psychiatry, 6, Toth, M., Flandreau, E. I., Deslauriers, J., Geyer, M. A., Mansuy, I. M.,
e702. https://doi.org/10.1038/tp.2015.200 Merlo Pich, E., & Risbrough, V. B. (2016). Overexpression of
Molet, J., Maras, P. M., Kinney‐Lang, E., Harris, N. G., Rashid, F., Ivy, A. S., forebrain CRH during early life increases trauma susceptibility in
Solodkin, A., Obenaus, A., & Baram, T. Z. (2016). MRI uncovers adulthood. Neuropsychopharmacology, 41(6), 1681–1690. https://
disrupted hippocampal microstructure that underlies memory doi.org/10.1038/npp.2015.338
impairments after early‐life adversity. Hippocampus, 26(12), Watson, D., & Clark, L. A. (1991). The Mood and Anxiety Symptom
1618–1632. https://doi.org/10.1002/hipo.22661 Questionnaire. Department of Psychology, University of Iowa.
Noroña‐Zhou, A. N., Morgan, A., Glynn, L. M., Sandman, C. A., Weathers, F. W., Blake, D. D., Schnurr, P. P., Kaloupek, D. G., Marx, B. P.,
Baram, T. Z., Stern, H. S., & Davis, E. P. (2020). Unpredictable & Keane, T. M. (2012). Clinician‐administered PTSD scale for DSM‐5.
maternal behavior is associated with a blunted infant cortisol National Center for PTSD.
response. Developmental Psychobiology, 62(6), 882–888. https://
doi.org/10.1002/dev.21964
Pechtel, P., & Pizzagalli, D. A. (2013). Disrupted reinforcement learning SUPP ORTING INFO RM ATION
and maladaptive behavior in women with a history of childhood
Additional supporting information can be found online in the
sexual abuse: A high‐density event‐related potential study. JAMA
Psychiatry, 70(5), 499–507. https://doi.org/10.1001/jamapsychiatry. Supporting Information section at the end of this article.
2013.728
Pittman, J. O. E., Floto, E., Lindamer, L., Baker, D. G., Lohr, J. B., & Afari, N.
(2017). VA escreening program: Technology to improve care for
post‐9/11 veterans. Psychological Services, 14(1), 23–33. https://doi. How to cite this article: Spadoni, A. D., Vinograd, M.,
org/10.1037/ser0000125 Cuccurazzu, B., Torres, K., Glynn, L. M., Davis, E. P., Baram, T.
Risbrough, V. B., Glynn, L. M., Davis, E. P., Sandman, C. A., Obenaus, A., Z., Baker, D. G., Nievergelt, C. M., & Risbrough, V. B. (2022).
Stern, H. S., Keator, D. B., Yassa, M. A., Baram, T. Z., & Baker, D. G.
Contribution of early‐life unpredictability to neuropsychiatric
(2018). Does anhedonia presage increased risk of posttraumatic stress
disorder? Current Topics in Behavioral Neuroscience, 38, 249–265.
symptom patterns in adulthood. Depression and Anxiety, 1–12.
Saunders, J. B., Aasland, O. G., Babor, T. F., de la Fuente, J. R., & Grant, M. https://doi.org/10.1002/da.23277
(1993). Development of the Alcohol Use Disorders Identification