The CARING Study: Examining Biological, Behavioral, and Genetic Mechanisms in the Intergenerational Transmission of Toxic Stress

Community Settings.

Families were recruited in partnership with community organizations that serve young families across the state, including public libraries, preschools, and YMCAs. Based on the preferences of the organization leadership, information about the study was distributed via flyers, email, or in-person by members of the research team. In-person events include family movie nights, community meals, and preschool parent information sessions. Flyers were available in both English and Spanish and included images that represented the racial/ethnic backgrounds of families at recruitment sites. Other community recruitment strategies included snowball recruitment and a collaboration with UConn KIDS, a program that maintains a listserv of Connecticut families interested in participating in research.

Hospital-based Setting.

We also recruited participants through the emergency department at Connecticut Children’s Medical Center, which has an infrastructure to support research recruitment. A research assistant (RA) reviewed electronic health records and assessed each child for study eligibility criteria. If the child appeared eligible and the child’s health status was designated as low acuity, the RA approached the family to describe the study, screen for eligibility, and provide take-home materials. The RA then scheduled a follow-up video call to obtain informed consent.

Adverse Childhood Experiences.

Caregivers completed the CDC-Kaiser ACEs scale, a retrospective measure that includes 10 dichotomous items (yes/no) to assess experiences of neglect, abuse, and household dysfunction prior to age 18 years (Felitti et al., 1998). The ACEs scale has been widely used and is associated with a myriad of poor health outcomes across adulthood (Petruccelli et al., 2019). Caregivers also completed the Childhood Trauma Questionnaire (CTQ), a 28-item retrospective measure that is specific to experiences of childhood maltreatment. Likert-scale items are calculated to create a total score and five subscales: physical abuse, emotional abuse, sexual abuse, physical neglect, and emotional neglect. The CTQ is reliable and valid across a wide range of samples and internal consistency was high in our earlier work (α =.83 - .96) (Bernstein et al., 2003; Condon et al., 2019c).

Benevolent Childhood Experiences.

Caregivers completed the Benevolent Childhood Experiences (BCE) scale, a retrospective measure of positive early life experiences in the first 18 years of life (Narayan et al., 2018). The BCE scale includes 10 dichotomous items (yes/no) and has high test-retest reliability (r=.80). The BCE scale is reliable and valid in individuals with marginalized identities and ACEs history (Merrick et al., 2019; Narayan et al., 2018).

Parental Reflective Functioning.

Caregivers completed the Parental Reflective Functioning Questionnaire (PRFQ), a self-report instrument with adequate internal consistency (α = .77–.85) designed to assess a mother’s interest in and ability to recognize her child’s mental states (Luyten et al., 2017). The PRFQ includes three subscales: interest and curiosity in mental states, certainty about mental states, and pre-mentalizing modes (Table 2).

Parenting Stress.

Caregivers completed the Parenting Stress Index-Short Form (PSI-SF), an instrument that is reliable and valid in ethnically diverse samples (Haskett et al., 2006; Reitman et al., 2002). The PSI-SF includes a total score and 3 subscales: parental distress (α = .75), difficult child (α= .82), and dysfunctional parent–child interactions (α= .85).

Child ACEs.

Using 9 items from the National Survey of Children’s Health, caregivers reported whether their child had experienced any of the following ACEs: parent/guardian death, parent/guardian divorce or separation, parent/guardian served time in jail, witnessed intimate partner violence, exposure to neighborhood violence, living with someone with mental illness, living with someone misusing alcohol or drugs, or was treated unfairly because of their racial/ethnic group (Walker et al., 2022).

Family Rest-Activity Patterns.

Wrist actigraphy data was used to assess the consistency of rest-activity patterns in the caregiver and child. Dyads wore a research actigraph (Philips Respironics Actiwatch 2) on their nondominant wrist for approximately 10 days. To aid interpretation, caregivers also completed a brief sleep diary that is emailed daily via Qualtrics (Qualtrics, 2005). Actigraphy is a reliable method for measuring sleep and rest-activity data and is validated against polysomnography in adults and preschool age children (Bélanger et al., 2013; Sadeh, 2011). Collection of dyadic actigraphy data was acceptable and feasible in our pilot study work (N=20 dyads; Table 3).

To assess rest-activity rhythms (RAR), we used individual actigraphy data and performed cosinor analysis using average activity counts over 10-minute epochs (via a SAS ^® macro) (Jeon et al., 2020). The intensity of RAR was measured by the ratio of the amplitude to the MESOR (Midline Estimating Statistic of Rhythm), a rhythm-adjusted mean for a 24 hour period, which is obtained from the cosinor analysis. Intra-daily variability (circadian rhythm fragmentation) and inter-daily stability (adherence to a RAR over consecutive days) were calculated using nonparametric methods (Zuurbier et al., 2015).

Caregiver and Child Sleep Health.

Actigraphy data was also used to objectively measure caregiver and child sleep. We used the Respironics Actiware^© software program (Philips, V.6.3) to estimate variables related to sleep health, including sleep duration, timing, and efficiency. To assess subjective sleep, caregivers completed the Children’s Sleep Habits Questionnaire-Abbreviated form (CSHQ-A), an instrument used to evaluate difficulty with bedtime, sleep behavior, and nighttime waking in young children (Owens et al., 2000).The CSHQ-A has adequate test-retest reliability (r=.62-.79) and is validated against wrist actigraphy in children (Markovich et al., 2015). Caregivers also reported on their own sleep health using PROMIS^® fatigue, sleep disturbance, and sleep-related impairment short forms (Cella et al., 2010). PROMIS measures are precise, reliable, and valid instruments developed as part of an NIH initiative to allow comparisons across populations (Broderick et al., 2013).

Family Routines.

Caregivers reported on routines using the Family Routines Inventory (FRI), a measure of strength-promoting family routines that occur in daily family life and are unrelated to race/social class (Jensen et al., 1983). The FRI has high test-retest reliability (r=.79), internal consistency (α =.78-82), and validity (Abar et al., 2017; Jensen et al., 1983).

Perceived Stress.

To assess general perceived stress, caregivers completed the Perceived Stress Scale, a Likert-scale instrument that is widely used, reliable, and valid (α =.78 - .91) (Cohen et al., 1994; Lee, 2012). To assess stress specific to the COVID-19 pandemic, caregivers also completed an adapted version of the reliable and valid Responses to Stress Questionnaire (Gargiulo et al., 2021; Stress and Coping Research Lab, 2020).

Physiologic Stress.

Caregivers’ stress was also evaluated using the biomarkers and health indicators described for children above: hair cortisol, salivary cytokines, salivary CRP, blood pressure, and BMI. These biomarkers are valid indicators of the physiologic stress response in adults (Granger & Taylor, 2020). Caregivers also reported their health history (e.g., chronic conditions) and health behaviors (e.g., cigarette/e-cigarette use).

Mental Health Symptoms.

Caregivers completed the PROMIS Anxiety scale, which evaluates fear, anxiety, hyperarousal, and somatic symptoms (e.g. worry, panic, dread, restlessness), and the PROMIS Depression scale, which evaluates negative mood and self-view, social cognition, and decreased positive affect and engagement (Cella et al., 2010). Both instruments are reliable and valid across diverse populations (Schalet et al., 2016). Caregivers reported on post-traumatic stress disorder (PTSD) symptoms using the PTSD Checklist-Civilian, a Likert-scale instrument that was highly reliable (α =.94) in our earlier work (Condon et al., 2019c; Weathers et al., 1993).

Racial Discrimination.

Caregivers completed the Experiences of Discrimination (EOD) scale, a reliable and valid instrument used to assess experiences of discrimination and harassment related to an individual’s racial or ethnic background (Krieger et al., 2005). The EOD includes nine items to evaluate the type and frequency of discrimination experienced in different settings (e.g. at work, getting housing, at a restaurant), and two items to assess responses to unfair treatment.

Neighborhood Environment.

Caregivers reported on their neighborhood environment using select subscales from an instrument developed by Muhajid et al. (2007). Likert-scale items are used to calculate subscales related to neighborhood safety, violence, social cohesion, and activities with neighbors. This instrument has high test-retest reliability (r=.60-.88) and is valid in diverse geographic populations (α =.73-.83) (Mujahid et al., 2007).

Social Support.

Caregivers completed two instruments to assess the presence of social support. The PROMIS Emotional Support scale assesses feelings of being cared for, being valued as an individual, and having confident relationships (Shensa et al., 2016). The PROMIS Instrumental Support scale assesses perceived availability of functional support, such as someone to help them if they are sick (Kent et al., 2020). Both instruments have high internal consistency (α=0.88-0.92; α=0.90, respectively) (Kent et al., 2020; Shensa et al., 2016).

Positive Affect.

Caregivers completed the PROMIS Positive Affect scale, a Likert-scale instrument that assesses momentary positive and rewarding experiences (e.g. pleasure, joy, contentment) (Cella et al., 2010). The scale is highly reliable (α=0.93) and is a valid indicator of mental health across a variety of clinical populations (Carlozzi et al., 2019).

Pre-processing and Storage.

All biomarkers were stored and analyzed at the UConn School of Nursing Biobehavioral Lab. Saliva samples for cytokine and c-reactive protein analyses were collected in Salimetrics SalivaBio Child’s Swabs (SCS; for child samples) and 2mL cryovial passive drool collection devices (for caregiver samples). All tubes were immediately frozen at −80°C. Child SCS tubes were aliquoted in 250uL volumes within 6 months of collection, then aliquots were stored at −80°C until day of analysis. Caregiver 2mL cryovials were frozen until the day of analysis, then a single 500uL aliquot was created for analysis.

Analysis.

Saliva was assessed for inflammatory markers using single- and multi-plex Ella kits designed by ProteinSimple for use on an Ella, a machine to run automated enzyme-linked immunosorbent assays (ELISAs), by Bio-techne. We quantified the concentrations of the cytokines IL-1β, IL-6, CXCL8/IL-8, TNF-α, and CRP following the procedure described by ProteinSimple. All cytokines were analyzed on a single panel, and CRP was quantified separately. For each cartridge, we prepared fresh 2-fold dilutions of saliva samples with sample diluent and then loaded each cartridge according to manufacturer’s instructions. All samples were run in triplicate, and standard curves and controls were prepared by the manufacturer and validated post-run. Each sample’s triplicate relative fluorescence unit (RFU) percent coefficient of variation (%CV) were evaluated post-run; any samples with %CV over 20% were rerun. Values that fall outside of the range of detection determined by the standard curve were discarded.

Pre-processing and Storage.

Hair samples from children and caregivers were stored in aluminum foil packets, with the scalp end clearly marked. Hair samples were stored at room temperature until cortisol is extracted.

Hair cortisol extraction and quantification.

We used a protocol for hair cortisol extraction based on the methods described in D’Anna-Hernandez et al. (2011) and Greff et al. (2019). Hair was first washed to remove any excess sweat or sebum from the external surface by first adding 5mL isopropanol to hair sample in a clean 15mL falcon tube, then shaking at 500rpm for 3 minutes at room temperature. After decanting the supernatant, we repeated for a second wash and let samples air dry for 3-4 days. Once dry, we weighed the samples and transfered 50mg of hair into a 2mL tube with Lysing Matrix I (MPBiomedicals, SKU 11691850-CF). We ground each sample using a FastPrep24 bead beater running at 6.5 m/s for 40 seconds for 2-3 rounds until hair was ground to a fine powder. We then added 1mL HPLC grade methanol to the ground hair and shake gently at room temperature for 24 hours. Immediately following this step, we centrifuged the sample/methanol solution for 2 minutes at 18,000g. Finally, we transferred 600uL of supernatant into a sterile 1.5 microcentrifuge tube and dry at 40C until evaporated. Once dry, we reconstituted each sample with 400uL of PBS pH 7.4 and stored at −80°C until ready for cortisol assay.

We quantified hair cortisol using a salivary cortisol kit by Salimetrics according to manufacturer’s instructions and interpreted plate results on a VersaMax microplate reader and accompanying SoftMax Pro 6.5 software. We evaluated the standard curve and unknown duplicates’ %CV and discarded any that exceeded 20%. After checking %CVs, we corrected all values to the plate blank to remove any background noise that was picked up by the plate reader. Using blank corrected values, we then determined the R² of the resulting standard curve (4-parameter logistic regression), and extrapolated the minimum value, maximum value, Hill’s slope, and point of inflection of the curve to obtain all unknown concentrations.

Pre-processing and Storage.

Saliva samples for DNA analyses were collected in ORAGENE OC-175 (for child samples) or OGR-500 (for caregiver samples) collection devices by DNA Genotek. These samples were stored at room temperature upon arrival for no more than 2 weeks before sample processing. Prior to aliquoting, saliva samples were gently inverted for five seconds, then incubated at 50°C for 2 hours. After incubation, the samples were aliquoted into 1.5 microtubes with 500uL per aliquot, then stored at −80°C until DNA extraction.

DNA Isolation.

We extracted DNA using the manual purification method recommended for Genotek OC-175 and OGR-500 devices (Document PD-PR-006 Issue 12/2018-10). The first aliquot of each sample was used for DNA extractions. Briefly, we thawed saliva aliquots and added 20uL of PT-L2P and incubate them on ice, followed by precipitation and purification with Ethanol. We quantified DNA concentration and purity for each sample via fluorescence method using a Biospectrometer. All DNA stocks were diluted to a final volume of 150uL and final concentration of 5ng/uL was used in SNP analyses.

SNP Genotyping.

We genotyped individuals for SNPs (Table 4) using TaqMan assays (40X) purchased from Applied Biosystem. All SNPs were run on a StepOnePlus Real Time PCR System machine from Applied Biosystems. On the day of each assay, each SNP was diluted to 20X with TE buffer pH 8.0, then mixed with PCR Genotyping Master Mix before pipetting into all wells. Diluted DNA was then delivered into each well with a multichannel pipette according to the Taqman SNP Genotyping protocol (Publication number MAN0009593, Revision B.0), totaling 15ng of DNA per well. All samples were run in duplicate for each genotyping assay, and negative controls were also included on each plate and run in duplicate. After the plate run, negative controls were verified and genotype calls were determined using Applied Biosystem StepOnePlus Software V2.3. If the software did not determine a call or if duplicates did not match, samples were rerun. If no call was confirmed on the second run, a new DNA extraction was created and run for a third time. Samples that remained uncalled by the third rerun were removed.