Participants
The 2004 Pelotas Birth Cohort is an ongoing population-based birth cohort study in southern Brazil that included all children born from mothers residing in the urban area of Pelotas municipality, RS, between January 1st and December 31st in the year 2004. Recruitment occurred by daily monitoring of all maternity hospitals in the city (about 99% of all births in Pelotas occur in hospital). Of all eligible children born in 2004, 99.2% (N = 4231) were included in the study (perinatal study). To date, nine assessments have been completed. First, mothers were interviewed within 24h postpartum using a standardized questionnaire. Follow-ups with mothers and their children occurred at 3 months, and at 1, 2, 4, 6, 11, 15, and 18 years of age. All follow-ups, except at age 15, had retention rates between 85.0–99.2%. The follow-up at age 15 started in November 2019 but ended abruptly in March 2020, when social distancing measures due to COVID-19 pandemic took place in Brazil, and data collection had to be interrupted, with a follow-up rate of 48.5% (N = 1949) of the original cohort [34]. Hair cortisol concentration (HCC) data were first collected from the cohort at age 15. While the follow-up rate at this wave was lower than earlier phases, the inclusion of these data adds significant value by enabling exploration of cortisol levels during adolescence, a critical developmental period. Further details can be found elsewhere [34–36].
At the perinatal study and in all follow-ups, structured questionnaires were used to obtain detailed information on demographic, socioeconomic, behavioral, mental health, and biological characteristics, as well as reproductive history, and health care utilization. The data collected were directly entered in the software REDCap (Research electronic data capture) [37].
In this study, the primary variables of interest were symptoms of mental health difficulties (the exposure) and hair cortisol concentration (the outcome). The analyses focused on three key follow-ups, at 6, 11 and 15-year years. Participants were included in the analysis if they had complete data on both mental health symptomatology in at least one of the ages and HCC.
Mental Health Symptomatology
Mental health symptomatology was assessed using the Strengths and Difficulties Questionnaire (SDQ) [38] at ages 6, 11 and 15. The SDQ has 25 items divided into five subscales, each with scores ranging from 0–10. The SDQ total difficulties score is the sum of the four sub-scales of inattention/hyperactive symptoms, conduct problems, emotional symptoms, and peer relationship problems (the 5th subscale, not used for the total score, is prosocial behavior). The SDQ was adapted and previously validated for the Brazilian population of children and teenagers aged between 4 and 16 years [39].
Hair Cortisol Concentration
Hair samples for cortisol concentration analysis were collected by trained personnel, following standardized methodology according to detailed instructions previously stablished [40]. Hair strands were cut from the vertex of the scalp, as close to the skin as possible. The 3 cm segment nearest to the scalp underwent processing steps of washing, pulverization, and weighing. Following these preparatory processes, cortisol was extracted, and its concentration quantified using established laboratory techniques for details, see [40]. After cortisol extraction, samples were suspended in 150 µl of assay diluent for 24 hours and then tested in duplicate using the ELISA technique with the High Sensitivity Salivary Cortisol Immunoassay Kit (Cat# 1-3002, Salimetrics, Pennsylvania). The ELISA Spectramax 190 plate reader was used for cortisol detection. The intra- and inter-assay variation coefficients were < 10%. Cortisol concentrations are expressed in pg/mg. Outliers, corresponding to values 4 SD away from the mean HCC using raw data, were removed from the analytical sample (N = 10). Due to the skewed distribution of HCC, data were transformed using a natural logarithm.
Covariates
Maternal and adolescent demographic and socioeconomic characteristics, anthropometric measures, harsh parenting exposure, hair characteristics, and the use of corticosteroid medications were used to describe the samples and as confounders in adjusted analyses. The confounders were selected from assessments prior to the age 15 follow-up, to avoid adjusting for mediators. The maternal characteristics measured during the perinatal study included family income (in quintiles), parity (1, 2, 3 or more children), age (≤ 19, 20–34, ≥ 35 years), and years of schooling (≤ 4, 5–8, ≥ 9 years). The adolescent characteristics collected at the perinatal study were sex (male/female), skin color - reported by the mother (white, black or brown, or other - yellow or indigenous), and birth weight (in grams, categorized as low weight, < 2500g, no/yes). Maternal depression was assessed at 12 months post-partum by the Edinburg Postnatal Depression Scale – EPDS (yes/no) using a cut-off score of 10 or more on the scale. Body mass index (BMI) for age in z-score (<-2SD, ≥-2SD to ≤ + 1SD, +1SD to ≤ + 2SD, >+2SD) was measured at ages 6 and 11. Regarding harsh parenting, the Parent-Child Conflict Tactics Scale (CTSPC) was used at 6 and 11 (Straus et al., 1998). The CTSPC is a 22-item questionnaire that measures parental behavior toward the child in the preceding 12 months. Four items about severe physical abuse were not administered in this study due to ethical reasons. We defined harsh parenting as comprising the sum scores of the psychological aggression (5 items; e.g., “Shouted, yelled, or screamed at him/her”), corporal punishment (5 items; e.g., “Spanked him/her on the bottom with our bare hands”), and physical maltreatment (4 items; e.g., “Slapped him/her on the face or head or ears”) subscales. All items were scored on a 3-point scale (never, once, or more than once), yielding a total score of 0 to 28. The following participant hair characteristics were measured at age 15: color (black, brown, blonde, or red), texture (straight, wavy, curly, or kinky), and weekly washing frequency (≤ 3, 4–6, ≥ 7 times), as well as medication use (the use of any corticosteroid in the 3 months prior to the interview – no/yes).
Statistical Analyses
All analyses were run in the statistical package Stata version 17.0 (College Station, TX: StataCorp LLC. StataCorp. 2017). Descriptive analyses involved calculation of absolute and relative frequencies, using the chi-square heterogeneity test for categorical variables and t test for continuous variables from participants (and mothers) included and not included in the analyses.
The key study associations, between SDQ scores and HCC (both examined as continuous variables) were estimated in crude and adjusted linear regression models, separately for each age: 6, 11 and 15 years. Beta values (β) and 95% confidence intervals (95%CI) were obtained. The β and CI values were exponentiated, considering that the outcome measure was inserted in the models in natural logarithm form. The Total SDQ scale and each subscale were tested in independent models. Confounders had been selected based on findings from previous studies on the topic and a theoretical model defined by the authors. Two adjustment models were estimated. Model 1 included family income at birth, maternal (age, education, and parity) and participants perinatal variables (sex, skin color, and weight at birth), maternal depressive symptoms at 12 months, hair variables (color, type, and weekly washing frequency), and corticoid use in the last three months. For mental health exposure at age 6, only model 1 was applied. The variables entered in Model 2 varied depending on the timing of mental health exposure. For exposure at age 11, BMI and the CTSPC harsh parenting score from age 6 were added to Model 1. For exposure at age 15, BMI at age 11 and the mean CTSPC score from ages 6 and 11 (to capture a more consistent pattern of harsh parenting prior to age 15) were added to Model 1.
Ethical Aspects
The Research Ethics Committee of the Faculty of Medicine of UFPEL approved the study protocol and all follow-ups of the Pelotas Birth Cohort of 2004. In all follow-ups, informed consent was obtained in writing from the mothers or legal guardians. At 15 years of age, the participants also signed an informed assent form.