Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

I appreciated the opportunity to review the manuscript presenting the first systematic review of research on non-suicidal self-injury (NSSI) conducted within the African context. The study aims to address the lack of African representation in global NSSI literature and explores the nature, risk factors, and functions of NSSI across 33 studies.

Overall, I found the manuscript to be well-written, with a strong command of the English language. The introduction included a historical overview of NSSI and framed the study within global and cross-cultural perspectives. Likewise, the study was conducted using well-established systematic review frameworks, including JBI, PRISMA, and MMAT, and the authors provided a clear, thorough description of the methodology.

Strengths

• A clear and concise introduction that summarized the potential contributions of the research from the African continent.
• Use of established systematic review frameworks (JBI, PRISMA, MMAT).
• Thorough description of the methodology used in the study, including supplements that show each step of the analysis process.
• Organization of the findings into relevant thematic categories, with tables summarizing the extant data.

Limitations

Despite its strengths, the manuscript does not fully address several critical questions that the authors raise in the introduction. Specifically, the analysis lacks exploration of:

• Gender and age differences in NSSI prevalence and functions.
• Regional, and in this study, income-level variations across African countries.
• Comparisons between Western and non-Western NSSI measures.
• Interactions or synergies among risk and protective factors that may highlight sociocultural distinctions.

By addressing the questions posed by the authors, the results can more fully highlight the unique contributions of the literature from the African continent to global NSSI research.

Recommendations:

• Reanalyze the “qualitised” data to examine gender, age, and income-level (regional) differences to better answer the questions proposed by the authors in the introduction.
• Compare findings between Western and non-Western measures to highlight sociocultural distinctions.
• Elaborate on the functions of NSSI and any distinctions by regional or sociocultural contexts.
• Expand the discussion of interactions/synergy among the risk and protective factors, possibly adding an additional column in Table 3 to highlight any regional or sociocultural distinctions.
• Provide a more in-depth interpretation of the findings from the African continent and their implications for global NSSI literature.

Author Response

REVIEWER 1

Comment 1: Reanalyze the “qualitised” data to examine gender, age, and income-level (regional) differences to better answer the questions proposed by the authors in the introduction.

Response 1: Thank you for this comment. We have added text to address this issue.

Revised text 1: Risk factors that were mentioned most often were a current or past history of mental health problems (16 studies, 48.5%) [19,25,67,68,77-80,82,84,86-88,90,91,94], a history of exposure to adverse childhood experiences (12 studies, 36.4%) [18,25,68,69,72,74,77,79,81,82,83,89]; and demographic factors including age and sex (11 studies, 33,3%) [18-20,25,73,75-77,79,83,90]. However, and contrary to expectations, country income level (lower-income, LMI, or UMI) did not emerge as a significant risk factor for NSSI engagement. (see p.18, highlighted in yellow in the text)

 

On a related theme: Further, the risk factors that were mentioned most often in reviewed studies (mental illness, exposure to ACES, age, and sex) correspond largely to findings from the extant literature [16,14,22,23,27,3037,38]. However, this review identified a number of socially mediated risk factors for NSSI – including discrimination, acculturation, and noncompliance with socially prescribed rituals [25,76,82,89] – that have not adequately been considered in the extant literature and which could, therefore, usefully be considered in future research. (see p.22, highlighted in yellow in the text)

 

Comment 2: Compare findings between Western and non-Western measures to highlight sociocultural distinctions.

Response 2: Thank you for this helpful comment. We agree and have added text to address the issue.

Revised text: At a broader level, studies in this review that assessed for NSSI using measures developed in Western countries have tended, not surprisingly, to produce findings that largely mirror Western conceptualisations of NSSI in terms of risk factors for, types of, and the functions of NSSI 1[8,19,72,78,91,93,94],  . However, exploratory studies that have employed broad open-ended questions (e.g., “Have you ever harmed yourself on purpose in a way that was not to take your life?”), that provide opportunities for participants to mention novel types of NSSI that are salient to them [21,68,71,74,82,84,87], have provided a broader socially contextualised conceptualisation of the nature and dynamics of NSSI that would appear to be more congruent with African emic understandings.  Finally, the one measure of NSSI that was identified in this review – the Self-Punishment Scale (SPS) [99] – that was developed and validated for use with African samples, assesses not only physical self-injury (as per the DSM), but also other forms of non-suicidal self-harm, including: (a) self-deprivation (e.g., “I deprive myself of sleep and food”), (b) thinking and affective self-harm (e.g., “I do things to make other people hate me”), and self-neglect (e.g., “I don’t care about my health”). This broader conceptualisation of NSSI would appear to reflect the fact that African conceptualisations of NSSI may be informed by different/emic understandings that  are not adequately captured by contemporary Western psychiatric conceptualisations. (p. 23, highlighted in red in the text)

 

Comment 3: Elaborate on the functions of NSSI and any distinctions by regional or sociocultural contexts.

Response 3: We believe that the specific functions of NSSI are comprehensively addressed in Table 4. Reviewed studies did not provide a clear indication of distinctions by regional or sociocultural context, suggesting the need for future research specifically designed to address this issue – with the need for such research being indicated in additional text added to the discussion section:

Additional text: Second, reviewed studies did not provide an adequate basis for exploring distinctions in the functions of NSSI across different regional and sociocultural contexts on the African continent, with there being a need for future research designed to specifically address this issue. (p. 24, highlighted in yellow in the text)

 

Comment 4: The analysis fails to explore interactions or synergies among risk and protective factors that may highlight sociocultural distinctions.

Response 4: The reviewer raises an important point. However, we would respectfully submit that this issue was not comprehensively addressed in the studies reviewed, with the only clear attempt to address the issue being provided by one or two papers. See, for example:

At a broader level, some studies found that NSSI outcomes were influenced by interactions or synergies involving both risk and salutary factors as well as influences emanating from different ecosystemic levels. For example, in the study conducted by Quarshie and colleagues [82] it was found that exposure to adverse events in the home involved influences emanating from: the intrapersonal level (e.g., the individual’s sex), the interpersonal level (e.g., punitive or abusive parenting styles), and the socio-cultural level (culturally defined perceptions regarding appropriate parenting practices). Such synergies are, of course, not particularly surprising in the context of contemporary conceptualizations of risk and resilience in terms of which the outcome of exposure to adverse life events has been found to be influenced by multisystemic transactions or synergies [100-103]. (p18. , highlighted in yellow in the text)

 

Further: …. with there being evidence that there may be synergies between certain risk and salutary influences. For example, in the study by Kok and colleagues [69] it was found that individuals who act mindfully are more likely to engage in self-harming behaviours (a risk factor) but less likely to engage in NSSI behaviours involving more serious injury (a protective factor). As such, a more comprehensive perspective on the interplay between NSSI risk and salutary influences would appear to be indicated. (p 22, highlighted in yellow in the text)

 

Comment 5: Provide a more in-depth interpretation of the findings from the African continent and their implications for global NSSI literature.

Response 5: We believe that by addressing comments 1-4 above we have been able to provide a more in-depth interpretation of study findings and the implications of study findings for global NSSI literature.

Reviewer 2 Report

Comments and Suggestions for Authors

This is an important synthesis that focuses on adolescent and young adult NSSI within African contexts, an underrepresented evidence base with clear clinical and public health relevance. The paper is already strong on conceptual framing and transparent reporting. My suggestions below focus on strengthening methodological clarity and drawing out practice and policy implications without overreaching beyond the available data.

• Clearly state up front the primary definition used for study selection (e.g., intentional self-inflicted injury without suicidal intent) and how you handled studies that used broader or ambiguous terms (e.g., “self-harm,” “deliberate self-harm,” or “suicidal and non-suicidal self-injury combined”). If you included studies with mixed constructs, specify the extraction rules (e.g., extracted non-suicidal subsamples only; contacted authors; or excluded where distinction was impossible).
• Because prevalence and function profiles are sensitive to thresholds (≥1 lifetime episode vs DSM-5 frequency), I recommend a short subsection explicitly mapping the thresholds used across studies and a brief sensitivity-style narrative. This will temper cross-study comparisons and highlight measurement heterogeneity.
• Your ecosystemic framing is a strength. Consider aligning your coding more explicitly with a widely used functions taxonomy (e.g., automatic–social × positive–negative reinforcement) and then show how context-specific sociocultural functions in African settings extend or nuance that taxonomy. A compact matrix (functions × exemplar quotes/behaviors) would make this very clear to readers.
• If you registered a protocol (e.g., PROSPERO/OSF), cite the registration and note any deviations. If not registered, add a sentence explaining why and include the protocol as supplementary material to enhance reproducibility.
• List every database (with exact platform vendor and date-range limits) and clarify whether you searched regional/grey sources (e.g., African Journals Online, theses, conference proceedings). Given regional publication patterns, a brief rationale for your grey literature approach and any language restrictions will help readers judge risk of publication bias.
• Provide 1–2 concrete examples illustrating how you applied key criteria in borderline cases (e.g., mixed suicidal/non-suicidal samples; school vs. clinical; migrant or diaspora populations). This increases confidence in screening reliability.
• You’ve done quality appraisal, great. Go one step further and show how it influenced interpretation. For example, indicate where high-risk-of-bias studies disproportionately drive particular findings (prevalence extremes, specific functions, or risk correlates). A brief “what changed when down-weighting low-quality studies” paragraph can be purely qualitative but is very informative.
• Explicitly quantify the proportion of convenience vs probability samples and discuss the implications for external validity and precision of prevalence estimates. If possible, stratify headline prevalence by sampling frame and setting (school, community, clinic).
• Provide a compact evidence map: number of studies by country, subregion, and World Bank income group; and identify notable “blank spots.” A small table or figure will help readers avoid overgeneralizing South Africa–centric evidence to the continent.
• Where data allow, distinguish early/mid/late adolescence from young adulthood; school-attending vs out-of-school youth; and clinical vs non-clinical samples. Even a brief narrative contrast will enrich interpretation.
• Summarize instruments used (e.g., DSHI, ISAS, single-item measures), the languages of administration, and any local validation evidence. If most tools were developed in HIC settings, call out potential construct undercoverage (e.g., culturally embedded practices that resemble NSSI but may be differently construed).
• Emphasize that cross-study prevalence differences likely reflect methodological variance (thresholds, recall window, sampling) at least as much as true geographic differences. A single, firm cautionary sentence can prevent misinterpretation.
• Your discussion of interpersonal, identity, and meaning-making functions is compelling. Strengthen it by including a few vivid, context-grounded exemplars to illustrate mechanisms such as stigma navigation, peer bonding, or regulation of culturally specific distress idioms.
• If the corpus allows, comment briefly on gender patterns (including boys/young men) and any data on sexual and gender minority youth. If data are scarce, explicitly flag this as a priority gap.
• Translate key findings into 2–3 practical takeaways for clinicians and school-based mental health providers in African contexts (e.g., screening feasible with short, locally adapted tools; ask explicitly about functions; integrate family/peer contexts in safety planning).
• If school-based samples dominate, discuss opportunities for universal and selective prevention (teacher training, peer-led programs, culturally grounded psychoeducation).
• Encourage the co-development of interventions with youth, families, and community leaders to ensure acceptability, address stigma, and align with local understandings of distress.

Author Response

REVIEWER 2

 

Comment 1: Clearly state up front the primary definition used for study selection (e.g., intentional self-inflicted injury without suicidal intent) and how you handled studies that used broader or ambiguous terms (e.g., “self-harm,” “deliberate self-harm,” or “suicidal and non-suicidal self-injury combined”).

Response 1: The definition of NSSI employed in this study rests on three requirements: (a) deliberate self-harm, (b) to the surface of the body, (c) in the absence of suicidal intent. Regardless of whether broader or more ambiguous terms were employed by authors to describe self-harming behaviours, studies were only included in this review if they reported on behaviours that met these definitional requirements. To further clarify the definition of NSSI used in this review, additional text has been added

Revised text: In order to ensure that the definition of NSSI used in the study was broad enough to encompass emic perceptions of NSSI, a broad definition of NSSI was employed: “Deliberate self-harm to the surface of the body in the absence of suicidal intent”. While adequately capturing the core defining characteristics of DSM criteria for NSSI, it is broad enough to encompass culturally specific notions of NSSI. For example, it has been documented that an individual cutting their own hair may be regarded as a form of NSSI by some members of the Australian aboriginal community [44,51]; with the definition of NSSI employed in this study being broad enough to accommodate such culturally specific understandings. (p. 5, highlighted in red in the text)

 

Comment 2: Because prevalence and function profiles are sensitive to thresholds (≥1 lifetime episode vs DSM-5 frequency), I recommend a short subsection explicitly mapping the thresholds used across studies and a brief sensitivity-style narrative.

Response 2: We agree and have added additional text to address this issue as best as we can:

Additional text: With regard to DSM frequency and duration requirements, only one study met DSM requirements for NSSI (at least five times in the past year) [18], with only one study meeting the duration requirement [18] and only nine studies meeting the duration requirement [18,19,25,76,82,89,90,92,94]. Such heterogeneity in measurement obviously impacts on the confidence with which cross-study comparisons can be made. (p. 16, highlighted in red in the text)

 

Comment 3: Consider aligning your coding more explicitly with a widely used functions taxonomy (e.g., automatic–social × positive–negative reinforcement) and then show how context-specific sociocultural functions in African settings extend or nuance that taxonomy. A compact matrix (functions × exemplar quotes/behaviours) would make this very clear to readers.

Response 3: Thank you for this comment. We believe that the data presented in Table 4 provides a clear taxonomic framework for NSSI functions based on reviewed studies, with the extension to the conceptual four factor model being reflected in the additional inclusion of socially mediated functions. These findings are additionally discussed in the context of the paper as follows:

The functions of NSSI were examined in 15 studies (39.4%) [25,68,72,74,76,79,80,82,83,86,89,92], with 10 studies (30.3%) [68,70-72,74,79,80,83,86,92] assessing intrapersonal functions, and five studies (15.2%) [69-71,79,82] assessing interpersonal functions. Taken together, these functions correspond to the types of NSSI functions described in the extant literature [57,58], with intrapersonal functions being designed to either cope with negative emotions/cognitions or to achieve a desired emotional/cognitive state and interpersonal functions being designed to cope with distressing interpersonal relationships or to obtain desired reactions from significant others. However, four studies (12.1%) [25,76,82,89] additionally considered sociocultural functions of NSSI, with sociocultural functions being designed to cope with socially mediated forms of distress or to obtain a desired sociocultural reaction. (pp. 22-23, highlighted in red in the text.

 

Comment 4. If you registered a protocol (e.g., PROSPERO/OSF), cite the registration and note any deviations. If not registered, add a sentence explaining why and include the protocol as supplementary material to enhance reproducibility.

Reply 4: We agree. The following sentence has been added to the text to clarify this matter.

Additional text 4: Although a research protocol was developed for the study (see supplementary data S7), this protocol was not registered with PROSPERO prior to the commencement of data collection.

 

Comment 5: List every database (with exact platform vendor and date-range limits) and clarify whether you searched regional/grey sources (e.g., African Journals Online, theses, conference proceedings). Given regional publication patterns, a brief rationale for your grey literature approach and any language restrictions will help readers judge risk of publication bias.

Reply 5. You raise some important issues – which we believe are adequately address under subheading 2.3 (p., highlighted in red in the text, and presented below):

Searches were conducted in six databases (PubMed, Scopus, PsychINFO, African Journals Online, African Index Medicus, and Sabinet African Journals), with searches being conducted from inception to 31 December 2024. The search strategy was informed by terms that emerged from the operationalisation of key constructs in the research question (Table 1), with the general form of searches being: research participants (or equivalent) AND the names of each African country AND non-suicidal self-injury (or equivalent) AND research design (or equivalent) AND data analysis (or equivalent) AND research type (quantitative, qualitative, or mixed method); with the syntax employed in all database searches being presented in supplementary data S1. In line with inclusion/exclusion criteria for the study, we included qualitative, quantitative, and mixed methods studies that reported original research in peer reviewed journals or in postgraduate theses/ dissertations on any date prior to 01 January 2025 and excluded conference proceedings/abstracts, reviews of the extant literature, editorials, and commentaries, with studies published in any language being considered for inclusion. (pp. 7-8, highlighted in red in the text)

 

Comment 6: Provide 1–2 concrete examples illustrating how you applied key criteria in borderline cases (e.g., mixed suicidal/non-suicidal samples; school vs. clinical; migrant or diaspora populations).

Reply 6: Thank you for this comment. Studies were included in the review if they reported on behaviours that met the study definition of NSSI, and if they qualified for inclusion based on this studies inclusion and exclusion criteria (i.e., regardless of the nature of the study sample or comorbidities).

 

Comment 7: You’ve done quality appraisal, great. Go one step further and show how it influenced interpretation. For example, indicate where high-risk-of-bias studies disproportionately drive particular findings (prevalence extremes, specific functions, or risk correlates).

Reply 7: Given that this review is essentially a qualitative review (relying exclusively on qualitative or de-qualitised’ data), it is not possible for us to meaningfully explore associations between variables  - e.g., study quality on the one hand and NSSI prevalence estimates, risk factors, or  functions on the other hand. [Although we believe that you raise an important point that needs to be addressed in future research].

 

Comment 8: Explicitly quantify the proportion of convenience vs probability samples and discuss the implications for external validity and precision of relevance valence estimates  and if possible, stratify headline prevalence by sampling frame and setting (school, community, clinic).

Reply 8: Thank you for raising this concern. The proportion of convenience vs probability samples is presented in Table 2, and summarised in the following additional text:

Additional text 8: On average, there were slightly more females than males in the study samples (Mfemale = 58.9%), with sampling strategies involving convenience sampling in 29 studies (87.9%) and probability sampling in four studies (12.1%), with this primary reliance on convenience samples serving to limit the generalisability of study findings. Among the four studies that reported probability sampling, three studies involved samples drawn from the general community, one study involved a school sample, with there being no studies that involved participants drawn from clinical settings. (p. 16, highlighted in red in the text)

 

Comment 9: Provide a compact evidence map: number of studies by country, subregion, and World Bank income group; and identify notable “blank spots.” A small table or figure will help readers avoid overgeneralizing South Africa–centric evidence to the continent

Reply 9: Thank you for this comment. A compact map is presented in Figure 2 (p. 15) with additional information requested being provided in the following text that appeared our original submission:

The data in Table 2 indicate that 33 studies were identified for this review, with these studies reporting on the experiences of 29,100 African adolescents and young adults.  In terms of the World Bank classification of countries by income [96]: 17 studies (51.5%) were conducted in one upper middle-income country (South Africa), 11 studies (33.3%) were conducted in lower middle-income countries (Egypt, Eswatini, Ghana, Kenya, Morocco, Nigeria, Tunisia), and 5 studies (15.2%) were conducted in low-income countries (Burkina Faso, Mali, South Sudan, Uganda). From Figure 2 it is evident that no studies were identified that reported on NSSI research conducted in the only high-income country in Africa (Seychelles), and no studies were identified that reported on NSSI research conducted in 42 out of 54 African countries. (p. 15, highlighted in red in the text).

 

Comment 10: Where data allow, distinguish early/mid/late adolescence from young adulthood; school-attending vs out-of-school youth; and clinical vs non-clinical samples. Even a brief narrative contrast will enrich interpretation.

Reply 10: Thank you for this suggestion, that we have addressed using the following text:

Adapted text 10:  In terms of age categories (adolescent = 10-19 years; young adult = 20-25 years) 16 studies employed exclusively adolescent samples [19,21,25,68-70,72-74,77,78,83-85,89,94], with a further 16 studies employing samples that contained both adolescents and young adults [18,20,67,71,75,79-82,86-88,90-92,93], and one study employing an exclusively young adult sample [76]. Sample sources included: clinical settings (n = 9, 27.3%), schools (n = 8, 24.2%), tertiary educational institutions (n = 7, 21.2%), and the general community (n = 6, 18.2%); with one study (3.0%) employing samples that included both community samples and school children and two studies (6.1%) employing combined samples of school and tertiary education students. (pp. 15-16, highlighted in red in the text)

 

Comment 11: Summarize instruments used (e.g., DSHI, ISAS, single-item measures), the languages of administration, and any local validation evidence. If most tools were developed in HIC settings, call out potential construct under coverage (e.g., culturally embedded practices that resemble NSSI but may be differently construed).

Reply 11: A summary of validated instruments used to assess for NSSI is provided in supplementary data S6. A cautionary note regarding instruments used is provided in the following text:

Original text 11: At a broader level, studies in this review that assessed for NSSI using measures developed in Western countries have tended, not surprisingly, to produce findings that largely mirror Western conceptualisations of NSSI in terms of risk factors for, types of, and the functions of NSSI 1[8,19,72,78,91,93,94],  . However, exploratory studies that have employed broad open-ended questions (e.g., “Have you ever harmed yourself on purpose in a way that was not to take your life?”), that provide opportunities for participants to mention novel types of NSSI that are salient to them [21,68,71,74,82,84,87], have provided a broader socially contextualised conceptualisation of the nature and dynamics of NSSI that would appear to be more congruent with African emic understandings.  Finally, the one measure of NSSI that was identified in this review – the Self-Punishment Scale (SPS) [99] – that was developed and validated for use with African samples, assesses not only physical self-injury (as per the DSM), but also other forms of non-suicidal self-harm, including: (a) self-deprivation (e.g., “I deprive myself of sleep and food”), (b) thinking and affective self-harm (e.g., “I do things to make other people hate me”), and self-neglect (e.g., “I don’t care about my health”). This broader conceptualisation of NSSI would appear to reflect the fact that African conceptualisations of NSSI may be informed by different/emic understandings that  are not adequately captured by contemporary Western psychiatric conceptualisations. (p. 23, second paragraph, highlighted in red)

 

Comment 12: Emphasize that cross-study prevalence differences likely reflect methodological variance (thresholds, recall window, sampling) at least as much as true geographic differences. A single, firm cautionary sentence can prevent misinterpretation.

Reply 12: Thank you for raising this important point, that is addressed in the following text:

New text: In sum, reviewed studies were characterised by marked heterogeneity (in relation to sampling, measurement, and design) that is likely to be reflected in methodological variance in study findings. (p. 16, highlighted in red in the text).

 

Comment 13: Your discussion of interpersonal, identity, and meaning-making functions is compelling. Strengthen it by including a few vivid, context-grounded exemplars to illustrate mechanisms such as stigma navigation, peer bonding, or regulation of culturally specific distress idioms.

Reply 13: We agree. Please see Table 4 for specific verbatim exemplars (p. 21)

 

Comments 14-16: These final three comments relate to interventions and prevention, with requests being made by the reviewer for additional comments by the authors regarding the implications of study findings for interventions and practice.

Reply 14-16: While we would be thrilled to be draw such conclusions, there are a couple of reasons why we are hesitant to do so. First, given the nature of the data obtained in this review, we are cautious about making recommendations (in relation to policy and prevention) that are not (or cannot be) justified by the data. Second, we feel that recommendations regarding optimal interventions and prevention strategies in the African context need to (ideally) be the product of collaborative engagement between researchers and community members, and not simply reflect the views of one or two academics.

Reviewer 3 Report

Comments and Suggestions for Authors

This is a very good article, and I applaud the authors’ care and attention to methodological issues. The writing is clear and the arguments are made well. There is a sound argument made for the review vis-à-vis most previous reviews having been conducted on data from other continents and omitting Africa, and good suggestions were made regarding adjusting frequency of NSSI occurrences.

I have no criticisms to make except that the DSM-V is actually the DSM-5, and I wonder why the authors did not refer to the more recent edition—the DSM-5-TR. Perhaps they can comment on that, or adjust their ms to use the later edition.

Author Response

REVIEWER 3

 

Comment 1: DSM-V is actually the DSM-5, and I wonder why the authors did not refer to the more recent edition—the DSM-5-TR

Response 1: Thank you for pointing this out. We agree. We have therefore changed DSM-V to DSM-5, and mentioned that NSSI was retained in DSM-5-TR.

Revised text: Although a proposal for a distinctive NSSI diagnosis was made as early as 1984 [3], it was only 30 years later that NSSI was included in the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) in 2013 [4], and retained in the more recent text revised edition of the DSM (DSM-5-TR) in 2022 [5]… (p. 2, highlighted in green in the text).