May 13, 2026
Adolescent Mental Health Outcomes Following Early Regulatory Problems
- Matic Kadiš1,
- Pauline W. Jansen2,3,
- Nicole Tsalacopoulos4,5,6,7,8,9,
- Rianne Kok3,
- Dieter Wolke6,9,
- Satja Mulej Bratec1
- 1Department of Psychology, Faculty of Arts, University of Maribor, Maribor, Slovenia;
- 2Department of Child & Adolescent Psychiatry/Psychology, Erasmus Medical Center (MC), University Medical Center Rotterdam, Rotterdam, the Netherlands;
- 3Department of Psychology, Education and Child Studies, Erasmus University Rotterdam, Rotterdam, the Netherlands;
- 4Department of Paediatrics, Monash Children’s Hospital, Monash University, Melbourne, Australia;
- 5Department of Pediatrics, University Hospital Würzburg, Würzburg, Germany;
- 6Department of Psychology, University of Warwick, Coventry, UK;
- 7School of Healthcare , University of Leicester, Leicester, UK;
- 8School of Psychological Sciences, Monash University, Melbourne, Victoria, Australia;
- 9Warwick Medical School, University of Warwick, Coventry, UK.
Protocol Citation: Matic Kadiš, Pauline W. Jansen, Nicole Tsalacopoulos, Rianne Kok, Dieter Wolke, Satja Mulej Bratec 2026. Adolescent Mental Health Outcomes Following Early Regulatory Problems. protocols.io https://dx.doi.org/10.17504/protocols.io.ewov1rk6plr2/v1
License: This is an open access protocol distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
Protocol status: Working
We use this protocol and it's working. While it does require to have the exact same datasets we used, any dataset requests from the GenR and ALSPAC cohorts should yield similar results, with possible differences due to withdrawn consent after the publication of this protocol.
Created: April 02, 2026
Last Modified: May 13, 2026
Protocol Integer ID: 314421
Keywords: Data harmonization, Longitudinal cohort, Mixed models, Multiple imputation, SPSS, Item-level mapping, CBCL, SDQ, Early regulatory problems, Adolescent mental health, adolescent mental health trajectory, adolescent mental health outcome, longitudinal linear mixed model, longitudinal associations between early infant, early regulatory problem, longitudinal association, child behavior checklist, following early regulatory problem, mental health domains from age, developmental stability, mental health dimension, early regulatory problems this protocol, difficulties questionnaire, regulatory problem, emotional problem, mental health domain, early rp, peer relationship problem, impact of early rp
Funders Acknowledgements:
Satja Mulej Bratec
Grant ID: C3360-24-452009
Dieter Wolke
Grant ID: EP/X023206/1
Disclaimer
DISCLAIMER – FOR INFORMATIONAL PURPOSES ONLY; USE AT YOUR OWN RISK
The protocol content here is for informational purposes only and does not constitute legal, medical, clinical, or safety advice, or otherwise; content added to protocols.io is not peer reviewed and may not have undergone a formal approval of any kind. Information presented in this protocol should not substitute for independent professional judgment, advice, diagnosis, or treatment. Any action you take or refrain from taking using or relying upon the information presented here is strictly at your own risk. You agree that neither the Company nor any of the authors, contributors, administrators, or anyone else associated with protocols.io, can be held responsible for your use of the information contained in or linked to this protocol or any of our Sites/Apps and Services.
Abstract
This protocol details the methodology for the paper "Adolescent Mental Health Outcomes Following Early Regulatory Problems," which investigates the longitudinal associations between early infant regulatory problems (RPs) and adolescent mental health trajectories. The syntax methodology follows a 7-step process. First, we create the RP composite, classifying participants into four groups: never, multiple, persistent, and multiple-persistent RPs. Next, we harmonize the Child Behavior Checklist (CBCL) and Strengths and Difficulties Questionnaire (SDQ) into four mental health dimensions: emotional problems, peer relationship problems, hyperactivity/inattention, and conduct problems. Finally, we use longitudinal linear mixed models to determine the impact of early RPs on these mental health domains from ages 10 to 18. These models also evaluate developmental stability and sex-specific patterns, supported by specific sensitivity analyses using unadjusted models and imputed covariates.
Guidelines
To use this protocol you need to have access to the ALSPAC and Generation R data outlined in the protocol.
The datasets analyzed during the current study are not publicly available due to data protection regulations and participant confidentiality. Data are available from the respective cohorts (the Avon Longitudinal Study of Parents and Children and the Generation R Study) upon reasonable request and formal approval of a research proposal by their respective executive and data access committees.
Materials
Syntaxes:
- 1_RP_ALSPAC.txt
- 1_RP_GenR.txt
- 2_ALSPAC_data_harmonization.txt
- 2_GenR_data_harmonization.txt
- 3_merge.txt
- 4_restructure.txt
- 5_main analyses.txt
- 6_imputation.txt
- 7_mixed_models_imputed.txt
ALSPAC:
- Participant identifiers.
- Covariate data: Sex, birth weight, gestational age, socioeconomic status, ethnicity, maternal prenatal depression, maternal prenatal anxiety, and age at assessment.
- Infant regulatory problem data: Parent-reported feeding, sleeping, and crying items across infancy and toddlerhood.
- Adolescent mental health data: Strengths and Difficulties Questionnaire (SDQ) items corresponding to emotional, peer, hyperactivity, and conduct problems at ages 9, 13 and 17.
GenR:
- Participant identifiers.
- Covariate data: Sex, birth weight, gestational age, socioeconomic status, ethnicity, maternal prenatal depression, maternal prenatal anxiety, and age at assessment.
- Infant regulatory problem data: Parent-reported feeding, sleeping, and crying items across infancy and toddlerhood.
- Adolescent mental health data: Child Behavior Checklist (CBCL) items corresponding to emotional, peer, hyperactivity, and conduct problems at ages 9-10, 13-14 and 17-18.
Software
- SPSS software.
Before start
ALL syntaxes must be amended to point to the working directory and relevant file paths before starting the analyses. It's best to have all files in one directory and point to it using the CD command.
Preparing the Data
Before executing the syntax sequence, configure your local environment and raw data files. Convert the uploaded syntaxes in .txt files to .sps files.
Directory Setup: All SPSS syntax files must be amended to point to your specific working directory and relevant local file paths before starting the analyses. It is recommended to consolidate all datasets and syntax files into a single directory.
Data Preparation: Begin with the raw ALSPAC and Generation R datasets containing the required variables (participant identifiers, early regulatory problem items, mental health questionnaire items, and covariates). Verify that the filenames of these raw datasets match the load commands within the initial syntax files (1_RP_GenR and 1_ALSPAC_RP_syntax), renaming the datasets or updating the syntax paths as necessary to ensure they load correctly. Be sure to recheck the ID variable names for your specific datasets in steps 1 and 2.
Construct the Regulatory Problems (RP) Composite
Run the 1_RP_GenR and 1_ALSPAC_RP_syntax syntaxes on the raw cohort data. This step processes parent-reported crying, sleeping, and feeding items across infancy and toddlerhood to derive the central exposure variable through the following sequence. The processed cohort datasets are saved as 1_RP_GenR.sav and 1_ALSPAC_RPs.sav.
Item Dichotomization and Indexing: Raw continuous and categorical items indicating RP symptoms frequencies are dichotomized and summed to create modality-specific index scores (crying, sleeping, feeding) at each developmental timepoint .
Domain Classification: Modality indices are re-dichotomized to establish the presence or absence of a specific RP domain at each age.
Group Classification: Participants are categorized into four mutually exclusive trajectory groups:
- Never RP: Zero RPs, or an isolated RP present at a single timepoint.
- Multiple RPs: Difficulties in two or more distinct domains occurring concurrently at a single timepoint.
- Persistent RPs: Difficulties in any domain persisting across three or more timepoints.
- Multiple-Persistent RPs: Difficulties meeting both the multiple and persistent criteria.
Extract and Harmonize the Outcomes
Run the 2_GenR_data_harmonization and 2_ALSPAC_data_harmonization syntaxes on their respective datasets. This step processes the raw covariate and adolescent outcome data to ensure cross-cohort comparability. The processed datasets are saved as 2_GenR_Harmonized.sav and 2_ALSPAC_Harmonized.sav.
Extracted Variables:
- ALSPAC: Covariates (sex, birth weight, gestational age, socioeconomic status, ethnicity, maternal prenatal depression, maternal prenatal anxiety, and age at assessment) and mental health outcomes from the Strengths and Difficulties Questionnaire (SDQ) at ages 9, 13, and 17.
- Generation R: Covariates (sex, birth weight, gestational age, socioeconomic status, ethnicity, maternal prenatal depression, maternal prenatal anxiety, and age at assessment) and mental health outcomes from the Child Behavior Checklist (CBCL) at ages 10, 14, and 18.
Recode and Rename: Extract, rename, and recode the specified covariate and outcome variables.
Z-standardization: Z-standardize maternal prenatal psychopathology (depression and anxiety) scores to align the scales.
Harmonization: Compute the four harmonized mental health domains (emotional problems, peer relationship problems, hyperactivity/inattention, and conduct problems) for each assessment age.
Merge Cohorts
Run the 3_merge syntax to pool the 2_GenR_Harmonized and 2_ALSPAC_Harmonized.sav datasets into a single combined wide-format dataset. This step generates a new sequential participant ID variable and computes a cohort-specific covariate (Generation R = 1; ALSPAC = 2) to control for between-cohort differences in subsequent models. The output dataset is saved as 3_Pooled_Wide.sav.
Convert to Long Format
Run the 4_restructure syntax on 3_Pooled_Wide.sav to recode the data from wide to long format to prepare for mixed models. Output dataset: 4_Pooled_Long.sav.
Run the Main Mixed Models
Run the 5_main_analyses syntax on the 4_Pooled_Long.sav dataset. This step executes three sets of mixed models for each of the four harmonized mental health domains. Note that effect sizes, such as Cohen’s d and partial eta-squared, must be calculated manually from the model outputs, using formulas provided.
Partial eta squared:
Figure 1: Formula for calculating pseudo partial eta-squared for omnibus effects in the mixed models, utilizing the F-statistic (F), numerator degrees of freedom (df1), and denominator degrees of freedom (df2).
Cohen's d:
Figure 2: Formula for calculating Cohen's d for pairwise group comparisons in the mixed models, where the pooled standard deviation is estimated utilizing the square root of the sum of the residual variance and the random intercept variance.
Main Adjusted Models: Tests the main effects of early RPs and assessment wave, including an RP-by-wave interaction to assess developmental stability. These models adjust for all covariates (e.g., sex, birth weight, gestational age, maternal psychopathology, socioeconomic status).
Unadjusted Sensitivity Models: Re-runs the primary models without covariates to confirm that the observed main effects of early RPs are robust.
Exploratory Models: Adds an RP-by-sex interaction term to the fully adjusted models to evaluate if the longitudinal associations differ between males and females.
Impute Missing Covariates and Run Sensitivity Models
Run the 6_imputation syntax on the 4_Pooled_Long.sav dataset. This step generates imputed datasets to account for missing covariate data. The generated data is saved as 5_Pooled_Long_Imputed.sav.
Mutiple imputation utilizes the Fully Conditional Specification (FCS) method with a linear scale model and no interactions, and generates 20 distinct imputations using a maximum of 10 iterations per imputation for variables sex, birth weight, gestational age, socioeconomic status, ethnicity, cohort, maternal depression and anxiety (both raw and Z-standardized), and age at assessment.
Run the 7_mixed_model_imputed syntax on the 5_Pooled_Long_Imputed.sav dataset. This step re-estimates the primary adjusted models to verify robustness against missing data bias.
Execute the main adjusted mixed models across the 20 imputed datasets for each of the four harmonized mental health domains to test the main effects of early regulatory problems, assessment wave, and their interaction, adjusted for all covariates . These models retain the Restricted Maximum Likelihood (REML) estimation, participant ID random intercept, and first-order autoregressive (AR1) covariance structure for repeated measures .
Extract the 20 individual p-values generated across the multiple imputation iterations for each target effect, and calculate the standard statistical median to report the final pooled p-value. This manual step is required because SPSS does not automatically pool F-values and specific pairwise comparisons for mixed models utilizing imputed data.
Protocol references
RP categorization follows the guidelines by Winsper & Wolke (2014), first used in the ALSPAC cohort:
Winsper, C., & Wolke, D. (2014). Infant and toddler crying, sleeping and feeding problems and trajectories of dysregulated behavior across childhood. Journal of Abnormal Child Psychology, 42(5), 831–843. https://doi.org/10.1007/s10802-013-9813-1
Harmonization of mental health domains, first used in practice in this article, has been developed by Baumann et al., (2024):
Baumann, N., Anderson, P. J., Johnson, S., Marlow, N., Wolke, D., & Jaekel, J. (2024). Harmonisation of assessments of attention, social, emotional, and behaviour problems using the Child Behavior Checklist and the Strengths and Difficulties Questionnaire. International Journal of Methods in Psychiatric Research, 33(1). https://doi.org/10.1002/mpr.2001
Acknowledgements
The authors would like to thank the study participants and their families, as well as all current and former group members, paediatricians, psychologists, and research nurses of the Avon Longitudinal Study of Parents and Children and the Generation R Study. The Generation R Study is conducted by the Erasmus MC, University Medical Center in close collaboration with the Erasmus University Rotterdam and the city of Rotterdam. The general design of the Generation R Study is made possible by long-term financial support from the Erasmus MC, University Medical Center, Rotterdam, the Netherlands, the Organization for Health Research and Development (ZonMw) and the Ministry of Health, Welfare and Sport.
This paper was developed as part of the work of the Infant2Adult project. Infant2Adult is an acronym of the project Neurobiological mechanisms of adverse mental health outcomes following early regulatory problems. The consortium members are Serena Defina, Barbara Gungl, Samuel Henry, Marina Horvat, Pauline Jansen, Matic Kadiš, Fatma Keskin Kržan, Sander Lamballais, René Mõttus, Ryan Muetzel, Satja Mulej Bratec (PI), Christian Sorg, Andero Uusberg, Dieter Wolke and Saša Zorjan. The authors have no competing or potential conflicts of interest.