Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Thank you for the opportunity to review this manuscript. I think this study adds real significance to the field and has clear potential to support policy and practice going forward. I've outlined my comments and suggestions below, which I hope will help the authors strengthen the manuscript further.

This manuscript looks at a Portuguese, culturally adapted, caregiver-report ACE measure based on the Finkelhor/Turner framework, in a community sample of 341 children aged 2 to 10. Three ACE scoring approaches are compared: an approximate original ACE score, a full 37-item ACE pool, and an adapted 14-item reduced Finkelhor/Turner score, against caregiver-reported socio-emotional outcomes, using correlations, regression, ROC analysis, domain-based analysis, and latent class analysis. This is an important question to be asking, given how much interest there is in the cross-cultural utility and developmental relevance of different ACE scoring approaches for younger children. One of the real strengths of the paper is that it doesn't just rely on a cumulative ACE count, but compares cumulative, domain-based, and person-centred approaches side by side. The finding that the approximate original ACE score outperformed both of the expanded scores, while the reduced Finkelhor score consistently did better than the full ACE pool, is genuinely interesting, and it's discussed well in relation to Turner et al. (2020). Overall the manuscript is well written and analytically thorough, and it makes a useful contribution to the ACE measurement literature. Most of what follows relates to transparency, interpretation, and balance rather than any fundamental issue with the design or analysis.

Missing data handling. The biggest thing I'd want clarified is how missing ACE items were handled when the scores were built. Participants were allowed to submit with unanswered items, and cases with a lot of missingness were later excluded, but it isn't stated whether missing ACE items were treated as zero, prorated, or handled some other way. Since the cumulative ACE scores are the main predictor variables here, this really needs spelling out, as different approaches could shift score distributions and the associations with outcomes.

Shared-method variance. Both ACE exposure and socio-emotional adjustment were reported by the same caregiver at the same time, which raises the possibility that some of the associations found are inflated by shared reporting tendencies rather than reflecting the true relationship. The manuscript touches on reporting bias in the limitations, but I think the implications of using a single informant for both variables deserves more explicit discussion, especially given that the paper's central conclusions are about comparative predictive performance.

The "approximate original ACE score." The authors are upfront that this isn't a full replication of the original ACE inventory, since emotional neglect wasn't available. But that caveat gets a bit lost given how strongly the superior performance of this score is framed elsewhere. The Abstract, Discussion and Conclusion at times read as though this establishes something about the original ACE approach itself, when really it's a nine-item approximation tested in one cultural context against one specific outcome. The finding stands, I just think it needs a slightly more cautious framing.

Multiple comparisons. There are a lot of tests being run here: multiple outcomes, three scoring approaches, domain models, ROC analyses, latent class comparisons. I'm not saying a formal correction like Bonferroni is necessary, especially given the exploratory nature of some of this, but the risk of inflated Type I error should be acknowledged somewhere, either in the analysis section or the limitations.

Practical significance of the differences. The authors do report ΔR², AIC/BIC, ROC indices and odds ratios alongside p-values, which I appreciate. But some of the differences between the reduced Finkelhor score and the full pool are fairly small in real terms, often just a few percentage points of explained variance. These are meaningful in a psychometric sense, but I think it's worth discussing more directly what these gains actually mean in practice, rather than focusing mainly on which score ranks where.

ROC findings and screening implications. The ROC analysis is handled thoughtfully, but one thing that really stands out is how low the positive predictive values are across all three scores, around .21 to .25. So while discrimination is acceptable, a lot of children flagged above the cut-off wouldn't actually meet the elevated SDQ threshold. This is discussed later on, but I think it deserves more prominence earlier, including in the Results and the Abstract, so readers don't come away with an overly optimistic sense of the screening utility here.

Discussion length. The Discussion is thorough and balanced, but it runs long, and a fair amount of it re-states results before getting to the interpretation. Tightening this up, particularly the sections on domain-level and latent class findings, would make the paper's broader argument land more clearly.

Overlap between the three ACE scores. Since all three scores are built from overlapping items, it would help to see the correlations between them reported directly. That would give readers a better sense of how much shared variance there is before comparing their predictive performance.

Safeguarding. The survey collected information on abuse, neglect, domestic violence, parental substance misuse, family mental illness and other serious adversities, through an anonymous online design. Anonymity protects participants, but it also means there's no way to follow up if something concerning comes up. I'd like to see a bit more on what support was offered to participants, such as signposting or distress protocols, and whether the inability to intervene in cases of possible risk was considered during ethical review.

Specific comments

Abstract: I'd soften the claim that the approximate original ACE score showed the "strongest overall predictive and discriminative performance." Adding a line noting the incomplete nature of the measure and the low PPVs across all three scores would balance this out.

Participants: A simple flow diagram (538 responses to 362 retained to 341 analytic sample) would make the sample construction easier to follow.

Procedure: The rationale for using the youngest eligible child makes sense, but this should also be flagged as a limitation, since it systematically excludes older siblings and could affect how representative the sample is within families. I'd also like more detail on safeguarding and support procedures given how sensitive the ACE items are.

Missing data: Please clarify exactly how missing ACE items were handled in the score calculations.

ACE scoring: Worth reporting the correlations between the three ACE scores.

Age range: Please justify the decision to keep 10-year-olds in the main analysis, given the developmental distinctions that underpin the Turner et al. framework. The sensitivity analysis excluding them is helpful, but the reasoning for the primary inclusion decision isn't fully explained.

Statistical analysis: Please state whether multiple comparisons were considered, and if so, how.

Table 2: The pattern is clear, but some of the differences between scores are fairly small. Worth commenting on the practical size of these differences, not just the ranking.

Table 3: The low PPVs deserve more emphasis in the Results, not just the Discussion.

Discussion: Generally strong and appropriately cautious, but would benefit from less repetition of results and more synthesis.

Limitations: Worth adding explicit mention of shared-method variance, the possibility of inflated Type I error from multiple testing, and a bit more caution around treating the approximate original ACE score as a benchmark, given it isn't the complete original inventory.

One additional thought. The manuscript already touches on resilience and protective factors, and notes that not every child exposed to ACEs goes on to show poor outcomes. I think this could be developed further by drawing on the growing literature on Positive Childhood Experiences (PCEs) and Benevolent Childhood Experiences (BCEs), which offer a useful counterbalance to a purely deficit-focused view of adversity and help explain why outcomes vary so much among children with similar ACE exposure. A short section distinguishing adversity exposure from protective and promotive experiences, such as supportive relationships, safety, and positive adult connections, would give readers a more rounded picture, and would also back up the authors' own point that ACE scores shouldn't be read as deterministic. Useful references here might include Narayan et al. (2018) on BCEs, Bethell et al. (2019) on PCEs, Bellis et al. (2017) on trusted adult support, and Crandall et al. (2019) on PCEs and adult mental health outcomes.

 

Author Response

Please find attached our response to the reviewers' comments.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Thank you for the opportunity to review the manuscript, “Measuring Adverse Childhood Experiences in Young Children: A Portuguese Caregiver-Report Study Comparing Conventional, Expanded, and Reduced ACE Scores.”

Introduction

The authors provided a good description of the original ACE study and the limitations with the original measure and the impact when this measure is used for screening or risk identification in children, as well as the issue with the cumulative scoring of the measure. Developmental timing is critical to ACEs and should be considered in the framework, as the author(s) have stated.  

Methods

Procedure

This section could be reduced. Stating that IRB approval and informed consent was obtained would be sufficient for some of the details in this section. Also stating that it was anonymous study would suffice without having to state that you did not collect IP addresses, provided contact information of the research team, etc.

Measures

Adverse Childhood Experiences. Please provide details about the Portuguese measure- such as the number of questions and the type of response (i.e., Likert scale, yes/no, etc.). Please provide reliability coefficients for this study and prior studies.

SDQ- please provide the number of questions for this measure, the number of questions to obtain the Total Difficulties scores, and reliability coefficients for the current study.

The manuscript could be strengthened if the authors described in more detail how the different ACEs scores were calculated in this section, as well as descriptions of each measure. They do this in the analysis section, but it needs to be clearer in the measures so that the reader has a good idea of how the questions are scored and being used in the analysis section. It will be important to discuss how many questions are in each of the ACE’s scores (Original ACE Score, Full ACE Pool Score, and Adapted reduced Finkelhor ACE Score), as well as how many questions are in common between the measures. It would be helpful to see a list of the questions for each measure. There is so much going on with the analysis that without solid foundational information about the measures, it is difficult to determine how similar or different the questions were for each score.

ACE Domains and SDQ Outcomes: The manuscript would be strengthened if the authors described more details about the theoretical domains used.

Discussion

The discussion section includes rich descriptions of the results. The manuscript would be strengthened if the researchers could provide headings throughout the discussion that align with headings in the results section.  

Author Response

Please find attached our response to the reviewers' comments.

Author Response File: Author Response.pdf