Mental Health and Social Isolation among Sub-Saharan African Migrants in Tunisia: A Cross-Sectional Quantitative Study

Dear Reviewer(s),

We would first like to thank you for your careful reading of our manuscript and for your constructive comments, which have helped meaningfully improve the rigor and clarity of this work.

Before responding point by point, we would like to flag an important element that explains a number of the changes you will notice between this version and the previous one. A thorough review of our dataset revealed transcription and data-handling errors made during the initial entry of the questionnaires, as well as an inconsistency between two variables affecting a number of responses. After correcting these errors and removing one incomplete record, the final dataset comprises 97 observations (down from 98 initially). All statistical analyses were fully recalculated on this corrected dataset. This work revealed that several results and conclusions presented in the first version were not supported by the actual data. These elements have been thoroughly revised in the attached manuscript, independently of the points raised by the reviewers, out of concern for scientific integrity.

Below are our detailed responses to each of your comments.

Reviewer comment	Response
Introduction — The BPS model is not convincingly applied to the mental health of migrants. Provide examples of biological factors and their interactions.	We have strengthened the presentation of Engel's model by providing concrete examples of biological factors, explicitly linked to the corresponding items in our anxiety scale, and clarified that this dimension was only measured indirectly — a limitation now explicitly acknowledged.
It is not stated where the various instruments were derived from.	We now specify that our scales were developed by the authors, drawing on the content of existing, publicly available questionnaires, but adapted into an original instrument tailored to this study.
The full scales should be provided in Appendices.	The complete set of 36 items is now provided in an appendix.
Measuring PTSD levels	We confirm that its measurement was not part of the original scope of the study; this choice is now explicitly acknowledged as a limitation, with a recommendation to use a validated instrument (PCL-5) in future research.
State which version of SPSS was used.	Specified (IBM SPSS Statistics, version 28, 2021).
Why was K-means selected? How were the profiles arrived at? What variables were put in the model?	The methodological paragraph has been substantially expanded: standardization of scores, justification of the choice of K-means, comparison of 2- to 6-cluster solutions, and the criteria used for the retained solution. The corrected data support a more stable three-profile solution rather than the four-profile solution reported initially.
Statistical power: mediation analyses likely underpowered — why were they carried out?	We confirm that these analyses had initially been considered but were ultimately not carried out, as the correlation that would have motivated them was not confirmed, and the sample did not provide sufficient power for this type of model. The residual reference to these analyses in the STROBE checklist has been removed.
The qualifying statement under Table 4 ("indicating a direct and robust link...") is neither necessary nor warranted.	This wording is no longer applicable, as the correlation it asserted is not confirmed by the corrected data; the note has been replaced with a straightforward, factual statement of the significant associations.
No formal ethics review was in place; an exemption statement should be cited.	We confirm that no formal ethics committee was accessible at the time of data collection. The ethics statement has been revised to state this explicitly, detailing the protective measures actually implemented, and this absence is now acknowledged as a limitation of the study.
Results — Is resilience the same as social resources? Terms used interchangeably (Table 2, Table 8), causing confusion.	A single term ("Social resources") is now used consistently throughout the manuscript, including tables and figures, and the Keywords list, which still referred to "resilience," has been updated accordingly.
Why wasn't the relationship between social isolation and anxiety measured? RQ2 indicates it should be.	This result, already present in our analyses, is now explicitly cited in the text.
over interpretation of  non-significant findings	The relevant statements have been reworded to avoid presenting an absence of statistical significance as proof of an absence of effect, with explicit mention of power limitations.
Similarly, rephrase "non-significant trends".	This wording has been corrected: the term "trend" applied to a non-significant coefficient has been replaced by "association" (Results, length-of-stay analysis).
Table 8: no consistency in how "symptoms" and "distress" are used.	The term "distress profiles" has been harmonized throughout, consistent with the fact that it is the distress dimensions, not the resources, that differentiate the clusters, including two remaining instances of "symptom profiles" in the Discussion and Conclusions that had been overlooked in the previous revision.
Severe (M=20) vs high (M=19) distress distinction seems arbitrary.	The three-cluster solution, which is more clearly differentiated, resolves this issue; the silhouette coefficient is reported for transparency regarding the limited robustness of cluster separation.
Not clear how age was included to support claims about younger vs older profiles.	The corresponding claim, unsupported by the analysis, has been removed from the manuscript, including from the Abstract.
The qualitative component is not well described and does not seem systematically implemented — advisable to remove it.	In line with your recommendation, the passages drawing on unsystematic field observations have been removed from the Discussion. A residual sentence in the Measures section, which still referred readers to such content in the Discussion, has also been removed for consistency.
Interpretation after Figure 1 should be moved to Discussion.	The interpretive paragraph has been moved to the Discussion; only a factual description of the clusters remains in Results.
Discussion & Limitations — "Social isolation and mental health" section requires strengthening.	Both this section and the male-vulnerability section have been entirely rewritten in light of the corrected data.
Gender differences in depression: reinstate that effect size was only moderate.	As this difference is no longer confirmed in the corrected data, the comment is no longer applicable; the effect sizes for the differences that are actually significant (isolation, acculturative stress) are reported in the text and corresponding table.
"Effects of length of stay": avoid reporting new analyses in Discussion.	This complementary analysis, initially reported for the first time in the Discussion, has been moved to follow the expected structure: announced in Methods, results reported in Results, with interpretation alone remaining in Discussion.
Interpretation of non-significance is problematic.	The relevant sentence has been reworded to avoid concluding an absence of effect from a non-significant result.
Clinical and policy implications could be elaborated and more context-specific.	This paragraph has been expanded with two recommendations directly grounded in our findings.

We hope these revisions fully address your expectations and remain available for any further clarification. We thank you again for the quality of your review, which has considerably improved the rigor of this work.

Sincerely,

Zyed Achour