Review Reports
- Hassan Abdourahman Awaleh 1,2,
- Tony Byamungu 3 and
- Jalila El Ati 5,6,*
- et al.
Reviewer 1: Anonymous Reviewer 2: Anonymous
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThere are a number of errors/omissions in this paper which need to be corrected.
- The abstract does not give any indication of the total sample size.
- Table 2 shows that total sample size as 2103 whereas the summation of boys and girls indicates 2067 – other tables suggest the correct total is 2067.
- Table 2 also shows the overall prevalences of stunting as 25.4 in boys and 21.2 in girls – whereas the summation of the various categories of stunting indicates prevalences of 30.7 and 27.8, respectively (which are in agreement with the numbers in Table 4).
- Table 2 also shows that stunting prevalences are not different between boys and girls whereas a chi-square test gives a value of 5.049 p = 0.025 which is significant. So the abstract and text need correcting.
- Tables 3, 4 and 5 do not present the sample sizes for region or occupation – a repetition of prevalences.
- The age categories show markedly different ranges – more sensible groupings might be <=12, 13-23, 24-35, 36-47 and 48-50.
- The analyses assume that only 1 child was sampled from each household – is this correct? If not, the reader should be told the numbers selected from each household as well as how the analyses have been adjusted to allow for 2 or more children sampled from each household as one might expect intra household similarity.
- Wealth Index – it is usual to present as quintiles – also why create a wealth index? when there is an International Wealth Index which has been used in many countries.
- Analyses – with crude analyses the overall p value should be given as well for each sub-category. Also a Bonferroni correction should be applied.
- Analyses – multiple regression. What the reader needs to know is which of the confounding variables best predict, for example, wasting. So a stepwise procedure should be used which removes the most significant variable first and so on until the p 0.05 limit is reached. Also, how well the overall model predicts, for example, wasting, and some measure of effect size.
- I would change the layout of the tables. Have a single table which presents the crude analyses for wasting, stunting, underweight and composite index – the category n’s will be the same so no repetition. Then have a new table which shows which of the variables best predict each of the 4 malnutrition measures in their order of inclusion. So when one compares wasting, stunting, underweight and composite index one can see if the same variables predict the 4 measures of malnutrition or not.
- An aspect which is not discussed is to what extent are the confounding variables independent. For example, do nomadic households tend to have less income, wealth index etc. – are income and wealth index correlated? Have the authors tested for multicollinearity?
Author Response
[Nutrients] Manuscript ID: nutrients-4095684
Title: Malnutrition among children under five in Djibouti: A Composite Index of Anthropometric Failure analysis from the 2023 multisectoral survey
|
Response to Reviewer 1 Comments |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
1. Summary |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Thank you very much for taking the time to review this manuscript and for your interest in our work and the interesting comments and important suggestions. Please find the detailed responses below and the corresponding revisions/corrections “red highlighted” in the re-submitted manuscript. We hope that all your points are adequately addressed. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
2. Questions for General Evaluation |
Reviewer’s Evaluation |
Response and Revisions |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Does the introduction provide sufficient background and include all relevant references? |
Yes |
Thank you very much for your evaluation. Please find the point-by-point response to your comments ans suggestions. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Is the research design appropriate? |
Must be improved |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Are the methods adequately described? |
Must be improved |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Are the results clearly presented? |
Must be improved |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Are the conclusions supported by the results? |
Must be improved |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Are all figures and tables clear and well-presented? |
Must be improved |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
3. Point-by-point response to Comments and Suggestions for Authors There are a number of errors/omissions in this paper which need to be corrected. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Comment 1: The abstract does not give any indication of the total sample size. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Response 1: We thank the reviewer for this observation. The total sample size has now been explicitly reported in the abstract. The added text reads: “a national random sample (n = 2103)” [page 1, line 26]. In addition, the Results section (3.1 Subjects) has been revised to clarify the analytical sample size. The following text was added: “Among the 2103 surveyed children, 36 were excluded due to absence at the time of measurement, missing information, or flagged anthropometric data; a total of 2067 children were included in the analyses” [page 6, lines 226–229]. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Comment 2: Table 2 shows that total sample size as 2103 whereas the summation of boys and girls indicates 2067 – other tables suggest the correct total is 2067. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Response 2: We thank the reviewer for highlighting this inconsistency. The value n = 2103 corresponds to the overall surveyed sample, whereas n = 2067 represents the final analytical sample after exclusion of 36 children due to absence at the time of measurement, missing information, or flagged anthropometric data. Table 2 has been corrected to report the analytical sample size (n = 2067), and consistency has been ensured across all tables and sections of the manuscript. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Comment 3: Table 2 also shows the overall prevalences of stunting as 25.4 in boys and 21.2 in girls – whereas the summation of the various categories of stunting indicates prevalences of 30.7 and 27.8, respectively (which are in agreement with the numbers in Table 4). |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Response 3: We thank the reviewer for identifying this inconsistency. The discrepancy was due to a transcription error in Table 2. The overall prevalence of stunting has been corrected to 30.6% among boys, 27.8% among girls, and 29.2% for both genders combined. These revised values are now consistent with the summation of the stunting categories presented in Table 2 and are in full agreement with the results reported in Table 4. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Comment 4: Table 2 also shows that stunting prevalences are not different between boys and girls whereas a chi-square test gives a value of 5.049 p = 0.025 which is significant. So the abstract and text need correcting. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Response 4: We thank the reviewer for this observation. The initially reported chi-square statistic and p-value were based on the erroneous stunting prevalence values in Table 2. After correcting the stunting prevalences, the chi-square test was recalculated, yielding a p-value of 0.16, indicating no statistically significant difference in stunting prevalence between boys and girls. The abstract and the relevant sections reflect now this corrected result. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Comment 5: Tables 3, 4 and 5 do not present the sample sizes for region or occupation – a repetition of prevalences. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Response 5: We thank the reviewer for pointing out this issue. In Tables 3, 4, and 5, prevalences were inadvertently reported in place of sample sizes for the variables region and occupation. This error has now been corrected, and all tables have been revised to clearly and accurately present both the sample sizes (n) and the corresponding prevalences. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Comment 6: The age categories show markedly different ranges – more sensible groupings might be <=12, 13-23, 24-35, 36-47 and 48-50. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Response 6: We thank the reviewer for this valuable suggestion. The age groupings used in our analyses (6–23, 24–47, and 48–59 months) were defined a priori based on standard epidemiological practice and their relevance to child growth and nutrition. The 6–23-month group corresponds to the critical period of complementary feeding and the first 1,000 days of life, during which growth faltering is most pronounced. The 24–47- and 48–59-month groups reflect subsequent developmental stages commonly used in national and international nutrition surveys to ensure comparability of results. While narrower age bands could provide greater granularity, the selected categories were chosen to maintain sufficient sample size within each group and adequate statistical power for multivariate multinomial logistic regression analyses. Using more disaggregated age categories would have resulted in smaller cell sizes for several outcomes, potentially compromising the robustness and stability of the estimates. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Comment 7: The analyses assume that only 1 child was sampled from each household – is this correct? If not, the reader should be told the numbers selected from each household as well as how the analyses have been adjusted to allow for 2 or more children sampled from each household as one might expect intra household similarity. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Response 7: We thank the reviewer for this important methodological point. For the survey, households were selected if they had at least one child under five years of age. In each selected household, all children aged 6–59 months were surveyed; therefore, more than one child could be included per household. To account for the non-independence of observations arising from multiple children within the same household, the analyses were conducted using survey-adjusted procedures that incorporated the household as the primary sampling unit, together with the sampling weights and stratification variables. This approach accounts for intra-household correlation and provides appropriate variance estimates. To clarified this point, we added the following text in the Methods section: In households with more than one eligible child, all children aged 6–59 months were included in the survey. Statistical analyses accounted for the clustered survey design, with households treated as the primary sampling unit. [page 3, lines 115-118]. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Comment 8: Wealth Index – it is usual to present as quintiles – also why create a wealth index? when there is an International Wealth Index which has been used in many countries. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Response 8: We thank the reviewer for this comment. A household wealth index was constructed as a proxy for long-term socioeconomic status as one of the indices used to assess household economic status. PCA-based wealth indices using housing characteristics and asset ownership are widely used in nutrition and health surveys, particularly in low- and middle-income settings. The index was derived from housing characteristics and context-specific household assets, including productive assets relevant to rural livelihoods. The first principal component, representing the wealth gradient, was retained. Households were categorized into tertiles (low, medium, high economic status) rather than quintiles to ensure adequate sample size and stable estimates in multivariable analyses. Although the International Wealth Index (IWI) is useful for cross-country comparisons, it relies on a fixed asset set that may not fully capture local economic realities. Therefore, a context-adapted PCA-based wealth index was considered more appropriate for within-country analyses. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Comment 9: Analyses – with crude analyses the overall p value should be given as well for each sub-category. Also a Bonferroni correction should be applied. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Response 9: Thank you for this important methodological comment. Firstly, overall p value for crude analysis were added in the tables 3,4,5,6 Regarding the use of the Bonferroni correction, we would like to present our point of view. In the univariate analysis, we used binary logistic regression models to assess the crude association between each explanatory variable and malnutrition. For categorical variables with more than two categories, odds ratios, 95% confidence intervals and p-values were reported for each category in comparison with a reference category, as is standard practice in regression-based analyses. In this context, the statistical significance of a categorical variable with multiple levels is appropriately assessed using a global test (e.g. likelihood ratio test or global Wald test), which evaluates whether at least one category differs from the reference. This global p-value corresponds to a single hypothesis test for the variable as a whole. Bonferroni correction is typically recommended when multiple independent hypothesis tests are performed simultaneously (for example, multiple pairwise comparisons between categories). In our analysis, no pairwise post-hoc comparisons were conducted; instead, each categorical variable was evaluated through a single global test within the regression framework. Therefore, applying a Bonferroni correction to the global p-value would not be statistically appropriate and could lead to an unnecessary loss of power. To improve clarity and address the reviewer’s concern, we have revised the Methods and Results sections to explicitly report the global p-value for each categorical variable in the univariate analysis, while retaining category-specific odds ratios relative to the reference group. [page 6, lines 199-221]. We hope this clarification addresses the concern and improves the transparency of the analysis. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Comment 10: Analyses – multiple regression. What the reader needs to know is which of the confounding variables best predict, for example, wasting. So a stepwise procedure should be used which removes the most significant variable first and so on until the p 0.05 limit is reached. Also, how well the overall model predicts, for example, wasting, and some measure of effect size. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Response 10: We thank the reviewer for this comment and the opportunity to clarify our analytical strategy. The aim of our multivariable logistic regression analysis was not to develop a predictive model but to estimate adjusted associations between for example stunting and a set of determinants that are well established in the literature. All included covariates (child’s gender; age in months, categorized as 6–23, 24–47, and 48–59 months; area of residence (urban/rural); household type (nomadic/ sedentary); region of residence; education level of the household head (low, middle, high); occupation of the household head (large-scale trader, worker/employee, senior or middle manager, unemployed/dependent on aid, small-scale trader); household wealth index (low, middle, high); household income index (low, middle, high); and household food security status assessed using the CARI index (severely food insecure, moderately food insecure, marginally food secure, food secure) were therefore selected a priori based on previous evidence and conceptual frameworks on child undernutrition, and were retained in the model regardless of their statistical significance. For this reason, we did not apply a stepwise variable selection procedure. Such data-driven approaches are primarily intended for predictive modeling and may lead to the exclusion of important confounders, potentially resulting in residual confounding and unstable parameter estimates. In etiological analyses, as in the present study, the use of a full model with theoretically justified covariates is considered more appropriate. Regarding model performance, the overall goodness-of-fit of the logistic regression model was assessed using the Hosmer–Lemeshow test, which indicated an adequate fit of the model to the data. The magnitude of associations was quantified using adjusted odds ratios with their 95% confidence intervals, which represent the effect sizes of the determinants examined. We have clarified these points in the 2.7 Statistical Analysis section of the revised manuscript. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Comment 11: I would change the layout of the tables. Have a single table which presents the crude analyses for wasting, stunting, underweight and composite index – the category n’s will be the same so no repetition. Then have a new table which shows which of the variables best predict each of the 4 malnutrition measures in their order of inclusion. So when one compares wasting, stunting, underweight and composite index one can see if the same variables predict the 4 measures of malnutrition or not. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Response 11: We thank the reviewer for this insightful suggestion aimed at improving comparability across malnutrition outcomes. We agree that a consolidated presentation of associated factors can be highly informative. However, due to the large number of explanatory variables included, combining all crude analyses into a single table would result in an excessively large and difficult-to-read table. As a compromise, we can add a new summary table that synthesizes the key associated factors retained in the final multivariate models for wasting, stunting, underweight, and the CIAF. This table allows direct comparison of whether the same variables predict different forms of malnutrition, while the full regression results are retained in separate tables for transparency and completeness. Table 7. Summary of key factors associated with child undernutrition across anthropometric indicators (adjusted analysis)
Values indicate direction of statistically significant associations in the final multivariate logistic regression models. ↑: higher odds of malnutrition outcome (p < 0.05); —: not statistically significant in adjusted models; ±: significance observed in some categories or attenuated after adjustment. Models were adjusted for child age and gender, household characteristics, socioeconomic factors, region, and food security status as described in the Methods section. CIAF: Composite Index of Anthropometric Failure. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Comment 12: An aspect which is not discussed is to what extent are the confounding variables independent. For example, do nomadic households tend to have less income, wealth index etc. – are income and wealth index correlated? Have the authors tested for multicollinearity? |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Response 12: We thank the reviewer for this important comment. Multicollinearity was not explicitly tested in the initial analysis; however, we have now assessed it using the Variance Inflation Factors (VIFs) derived from an auxiliary linear regression model including the same covariates. The results did not indicate problematic multicollinearity, and the main findings of the logistic regression models remained unchanged. On the other hand, the confidence intervals of the odds ratios were narrowed for all variables. We clarify this point in the 2.7 Statistical Analysis section. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
4. Response to Comments on the Quality of English Language |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Comment: The English is fine and does not require any improvement |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Response: Thank to the reviewer for this appreciation. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Author Response File:
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsThis is an interesting and important secondary data analysis given the little previous information on childhood nutritional status in Djibouti. The report is well presented. However, I consider that there are a few issues that should be addressed/clarified:
Main concerns
Table 1 appears incorrect as WAZ should be used to classify underweight and WHZ for wasting. It is not clear whether or not this apparent misclassification is carried forward to the remaining analyses (e.g. Table 2). Also, the labelling of the CIAF categories (A-Y) differs between tables 1 and 2. Please harmonise.
The methods section (lines 194-199) should clarify how potential confounders were included in the multivariate multinomial logistic regression and what these were. This is especially important given the unexpected absence of associations between undernutrition and socioeconomic indicators and household food security.
Much of the discussion section covers the variables associated with undernutrition (e.g. age, sex). This is valuable (given the limited previous research in Djibouti), but these findings could be covered briefly. The main emphasis should be on how the CIAF provided greater or different insights than the separate anthropometric indices. In other words, sections 4.2 to 4.4 should justify using the CIAF in addition to the separate anthropometric indices to provide support for the statements in section 4.1. Specifically, how would nutrition policy in Djibouti change based on the CIAF in addition to the separate indices?
Under study limitations, it would be useful to justify why overweight/obesity was not included in the CIAF as this has been done in some previous reports. Similarly, there is no mention of micronutrient malnutrition which would provide a more comprehensive assessment of nutritional status.
Other comments
Lines 182-3: “Economic status was assessed using two indices, a wealth index and an income index...” is unclear. Please clarify how this was done (and cite references if standard instruments used).
Line 185: cite reference for “CARI”.
Table 2 reports n=1063 boys and n=1004 girls but n=2103 (rather than 2067) for both sexes. Please clarify. If due to missing data for sex, please note this in the table legend. Also, for the CIAF categories (A to F), does P=0.02 refer to all categories – rather than just category A? Please clarify. Finally, would it be possible to illustrate the overlap between the 3 undernutrition categories using a Venn diagram?
The text in the results section should be limited to reporting findings (and their interpretation saved for the discussion section). E.g. lines 205, 210 and 220: Sentences such as “Malnutrition remains a major public health concern among children aged 6–59 months in Djibouti”; “This broader measure underscores that conventional indicators, when used in isolation, may underestimate the true magnitude of the malnutrition burden” and “These findings suggest 220 that while traditional indicators provide valuable information on specific forms of malnutrition, the CIAF offers a more comprehensive picture of the overall nutritional vulnerability of children under five in Djibouti.” would be better moved to the discussion section. The results section could be shortened considerably.
Lines 216-220 do not present data according to gender: suggest move to 3.1.1
Please use either sex or gender consistently throughout the report (or justify use of both terms).
Lines 459-462 should clarify how other researchers and stakeholders could access the data used in this study.
Line 452: Should the University of Senghor (Alexandria, Egypt) be included in the funding statement?
Author Response
[Nutrients] Manuscript ID: nutrients-4095684
Title: Malnutrition among children under five in Djibouti: A Composite Index of Anthropometric Failure analysis from the 2023 multisectoral survey
|
Response to Reviewer 2 Comments |
||
|
1. Summary |
||
|
We thank the reviewer for these constructive comments. The manuscript has been carefully revised to improve the clarity and completeness of the Methods section, with additional details provided on the study design, data collection, and analysis. The Results section has been restructured and clarified to enhance readability and coherence. The Conclusions have been revised to ensure they are fully supported by the presented results. In addition, all figures and tables have been reviewed and improved to enhance clarity and presentation. |
||
|
|
||
|
2. Questions for General Evaluation |
Reviewer’s Evaluation |
Response and Revisions |
|
Does the introduction provide sufficient background and include all relevant references? |
Yes |
We thank the reviewer for his/her appreciation. |
|
Is the research design appropriate? |
Yes |
|
|
Are the methods adequately described? |
Must be improved |
Thank you very much for taking the time to review this manuscript. Please find the detailed responses below and the corresponding revisions/corrections highlighted in red in the re-submitted file. |
|
Are the results clearly presented? |
Must be improved |
|
|
Are the conclusions supported by the results? |
Must be improved |
|
|
Are all figures and tables clear and well-presented? |
Can be improved |
|
|
|
||
|
3. Point-by-point response to Comments and Suggestions for Authors |
||
|
Main concerns |
||
|
Comments 1: This is an interesting and important secondary data analysis given the little previous information on childhood nutritional status in Djibouti. The report is well presented. However, I consider that there are a few issues that should be addressed/clarified. |
||
|
Response 1: We thank the reviewer for this comment. We hope that the issues are adequately addressed in the revised manuscript. |
||
|
Comments 2: Table 1 appears incorrect as WAZ should be used to classify underweight and WHZ for wasting. It is not clear whether or not this apparent misclassification is carried forward to the remaining analyses (e.g. Table 2). Also, the labelling of the CIAF categories (A-Y) differs between tables 1 and 2. Please harmonise. |
||
|
Response 2: We agree with this relevant comment. We have, accordingly correct the Table 1 and harmonised the Table 1 and the Table 2. |
||
|
Comment 3: The methods section (lines 194-199) should clarify how potential confounders were included in the multivariate multinomial logistic regression and what these were. This is especially important given the unexpected absence of associations between undernutrition and socioeconomic indicators and household food security. |
||
|
Response 3: Potential confounders were identified a priori based on an extensive review of the literature and their documented associations with child nutritional status and socioeconomic conditions. All selected variables were entered simultaneously into the multivariate multinomial logistic regression model to control for confounding effects. The covariates included were: child’s gender; age in months, categorized as 6–23, 24–47, and 48–59 months; area of residence (urban/rural); household type (nomadic/sedentary); region of residence; education level of the household head (low, middle, high); occupation of the household head (large-scale trader, worker/employee, senior or middle manager, unemployed/dependent on aid, small-scale trader); household wealth index (low, middle, high); household income index (low, middle, high); and household food security status assessed using the CARI index (severely food insecure, moderately food insecure, marginally food secure, food secure). The additional text highlighted in red in the Methods section reads: Potential confounders were identified a priori based on the literature and included child’s gender; age in months, categorized as 6–23, 24–47, and 48–59 months; area of residence (urban/rural); household type (nomadic/sedentary); region of residence; education level of the household head (low, middle, high); occupation of the household head (large-scale trader, worker/employee, senior or middle manager, unemployed/dependent on aid, small-scale trader); household wealth index (low, middle, high); household income index (low, middle, high); and household food security status assessed using the CARI index (severely food insecure, moderately food insecure, marginally food secure, food secure). These variables were entered simultaneously into the multivariate models to control for confounding. [page 6, lines 206-215]. |
||
|
Comment 4: Much of the discussion section covers the variables associated with undernutrition (e.g. age, sex). This is valuable (given the limited previous research in Djibouti), but these findings could be covered briefly. The main emphasis should be on how the CIAF provided greater or different insights than the separate anthropometric indices. In other words, sections 4.2 to 4.4 should justify using the CIAF in addition to the separate anthropometric indices to provide support for the statements in section 4.1. Specifically, how would nutrition policy in Djibouti change based on the CIAF in addition to the separate indices? |
||
|
Response 4: We thank the reviewer for this important and constructive comment. We agree that the principal contribution of this study lies in demonstrating the added value of the CIAF beyond conventional anthropometric indicators. In response, we have streamlined Sections 4.2 to 4.4 by shortening descriptive comparisons related to age and gender and refocusing the discussion on how the CIAF provides complementary and policy-relevant insights compared with wasting, stunting, and underweight considered separately. Specifically, the revised Discussion (highlighted in red) now emphasizes how the CIAF captures children experiencing multiple concurrent anthropometric failures, identifies high-risk subpopulations that may be underestimated by single indicators, and supports more integrated and targeted nutrition policy responses in Djibouti. We have also added explicit discussion on how the use of the CIAF could inform national nutrition surveillance, programme prioritization, and resource allocation, thereby strengthening the justification for its use alongside conventional indicators. |
||
|
Comment 5: Under study limitations, it would be useful to justify why overweight/obesity was not included in the CIAF as this has been done in some previous reports. Similarly, there is no mention of micronutrient malnutrition which would provide a more comprehensive assessment of nutritional status. |
||
|
Response 5: We thank the reviewer for this pertinent comment. In this study, the CIAF was restricted to undernutrition indicators (stunting, wasting, and underweight) in accordance with its original conceptualization, which aims to capture anthropometric failure related to growth deficits, a public health problem in Djibouti. Although some recent reports have extended the CIAF to include overweight/obesity, there is currently no universally accepted definition or standardization of an “extended CIAF,” which may limit comparability across studies. Regarding micronutrient malnutrition, we fully agree that its inclusion would provide a more comprehensive assessment of nutritional status. However, the 2023 survey database used for the present analyses did not include biochemical or clinical indicators of micronutrient status, precluding their assessment. This limitation has now been explicitly acknowledged in the revised manuscript, and we highlight the need for future surveys to integrate micronutrient data to better capture the multiple dimensions of malnutrition. The added text in 4.6. Strengths and Limitations reads: In addition, micronutrient malnutrition could not be assessed because the 2023 survey database did not include biochemical or clinical indicators of micronutrient status. [page 17, lines 482-484]. Finally, overweight and obesity were not included in the CIAF, as the index was applied in its original formulation focusing on anthropometric failure related to growth deficits, and no standardized definition of an extended CIAF including overnutrition has yet been universally adopted, which supports comparability with existing literature. [page 17, lines 486-490]. |
||
|
Other comments |
||
|
Comment 6: Lines 182-3: “Economic status was assessed using two indices, a wealth index and an income index...” is unclear. Please clarify how this was done (and cite references if standard instruments used). |
||
|
Response 6: We thank the reviewer for requesting clarification. Household economic status was assessed using two complementary indicators: a wealth index and an income index, which capture different dimensions of socioeconomic position. Wealth index: a proxy indicator of household economic status was constructed using an integrated analysis of housing characteristics and household assets. Housing-related variables included type of dwelling, number of persons per room, source of lighting, source of drinking water, and type of cooking fuel. Asset ownership was assessed through 15 items: bed, mattress, table/chairs, radio, television, satellite dish, mobile phone, wheelbarrow, refrigerator/freezer, iron, mosquito net, fishing tools, boat, canoe, and agricultural tools. Principal component analysis (PCA) was applied, and the first principal component, representing a gradient of household wealth, was retained as the economic status index. Households were subsequently classified into low, medium, and high economic levels based on tertiles of the wealth index distribution [31-33]. Income index: current household economic status was assessed using a monthly income index. Total household income self-reported for the month preceding the survey was ranked and classified into low, middle, and high income categories according to tertiles of the income distribution. The Methods section has been revised to clearly describe the construction and rationale of both indices, with appropriate references added. [page 5, lines 184-191] 31. Traissac, P.; Martin-Prevel, Y. Alternatives to principal components analysis to derive asset-based indices to measure socio-economic position in low- and middle-income countries: the case for multiple correspondence analysis. Int J Epidemiol 2012, 41, 1207-1208. 32. Gartner, A; El Ati, J; Traissac, P; Bour, A; Berger, J; Landais, E; El Hsaïni, H; Ben Rayana, C; Delpeuch, F. A double burden of overall or central adiposity and anemia or iron deficiency is prevalent but with little socioeconomic patterning among Moroccan and Tunisian urban women. J Nutr 2014, 144, 87-97. 33. Traissac, P; El Ati, J; Gartner, A; Ben Gharbia, H; Delpeuch, F. Gender inequalities in excess adiposity and anaemia combine in a large double burden of malnutrition gap detrimental to women in an urban area in North Africa. Public Health Nutr 2016 19,1428-37.
|
||
|
Comment 7: Line 185: cite reference for “CARI”. |
||
|
Response 7: As requested, we added the reference for “CARI”: 34. World Food Programme (WFP). Technical guidance for WFP’s consolidated approach for reporting Indicators of food security (CARI). United Nations World Food Programme. Food Security Analysis (VAM) Rome, Italy. Third Edition December 2021. [page 5, line 194 & page 21, lines 616-617]
|
||
|
Comment 8: Table 2 reports n=1063 boys and n=1004 girls but n=2103 (rather than 2067) for both sexes. Please clarify. If due to missing data for sex, please note this in the table legend. Also, for the CIAF categories (A to F), does P=0.02 refer to all categories – rather than just category A? Please clarify. Finally, would it be possible to illustrate the overlap between the 3 undernutrition categories using a Venn diagram? |
||
|
Response 8: We thank the reviewer for drawing our attention to this inconsistency. The value n = 2103 corresponds to the overall surveyed sample, whereas n = 2067 represents the final analytical sample after excluding 36 children due to absence at the time of measurement, missing information, or flagged anthropometric data. The value reported in Table 2 has been corrected accordingly, and the correct total sample size is n = 2067. Regarding p=0.02, it is for all CIAF categories and it was also corrected in the table (p=0.06). Regarding the helpful suggestion to illustrate the overlap between the wasting, stunting, and underweight, we have added a Venn diagram showing the distribution and coexistence of the three forms of undernutrition among children 6-59 months. This figure visually complements the CIAF classification presented in Table 2 and highlights the proportion of children experiencing multiple concurrent anthropometric failures. The Venn diagram has been included in the Results section (Section 3.2) as a new figure (Figure 1), and is referenced accordingly in the text.
Added in the text: Figure 1. Overlap between wasting, stunting, and underweight among children aged 6–59 months in Djibouti (n = 2067) Legend: Venn diagram illustrating the distribution and overlap of the three conventional anthropometric indicators of undernutrition: wasting (weight-for-height z-score < −2 SD), stunting (height-for-age z-score < −2 SD), and underweight (weight-for-age z-score < −2 SD), based on WHO Child Growth Standards. Values represent weighted prevalence (%) of children experiencing single or multiple anthropometric failures. The diagram highlights the coexistence of undernutrition forms and complements the Composite Index of Anthropometric Failure (CIAF) classification. |
||
|
Comment 9: The text in the results section should be limited to reporting findings (and their interpretation saved for the discussion section). E.g. lines 205, 210 and 220: Sentences such as “Malnutrition remains a major public health concern among children aged 6–59 months in Djibouti”; “This broader measure underscores that conventional indicators, when used in isolation, may underestimate the true magnitude of the malnutrition burden” and “These findings suggest that while traditional indicators provide valuable information on specific forms of malnutrition, the CIAF offers a more comprehensive picture of the overall nutritional vulnerability of children under five in Djibouti.” would be better moved to the discussion section. The results section could be shortened considerably. |
||
|
Response 9: We thank the reviewer for this helpful suggestion. The Results section has been carefully revised to ensure that it is strictly limited to the reporting of empirical findings. All interpretative and contextual statements, including those highlighted by the reviewer have been removed from the Results section and relocated to the Discussion. As a result, the Results section has been substantially streamlined and shortened to improve clarity and compliance with the journal’s reporting standards. The removed text has been relocated in Discussion text (highlighted in red) so it fully absorbs and builds on those interpretative statements. |
||
|
Comment 10: Lines 216-220 do not present data according to gender: suggest move to 3.1.1 |
||
|
Response 10: We thank the reviewer for this observation. The text previously presented been revised to ensure consistency with the structure of the Results section. Gender-specific analyses are now reported exclusively in Section 3.2.2 (Gender differences in malnutrition), where all data are systematically disaggregated by gender. |
||
|
Comment 11: Please use either sex or gender consistently throughout the report (or justify use of both terms). |
||
|
Response 11: We harmonized all the text using gender term. |
||
|
Comment 12: Lines 459-462 should clarify how other researchers and stakeholders could access the data used in this study. |
||
|
Response 12: The data from this study are jointly owned by UNICEF, WFP and the Ministry of Health of Djibouti. Researchers and stakeholders may request access the data used in this study by sending an official request to the Ministry of Health, Djibouti: Programme National de Nutrition – Ministère de la Santé, Djibouti. E-Mail: prognutrition.dj@gmail.com Phone: +253 21340309 Web site: www.sante.gouv.dj |
||
|
Comment 13: Line 452: Should the University of Senghor (Alexandria, Egypt) be included in the funding statement? |
||
|
Response 13: We thank the reviewer for this important clarification. The University of Senghor (Alexandria, Egypt) did not provide financial or logistical support for the design, implementation, or analysis of this study. The 2023 national survey in Djibouti was conducted by the Ministry of Health and financially supported by UNICEF and the World Food Programme (WFP). The mention of the University of Senghor in the Acknowledgements was solely related to the academic affiliation of one author (Mr. Awaleh), who is a doctoral student at Senghor, and to the fact that the University may cover the article processing charges in the event of publication. To avoid any ambiguity, the Funding statement has been revised to include only UNICEF and WFP, and the Acknowledgements section has been amended to clearly reflect the non-financial nature of Senghor’s contribution. The amended text reads: Acknowledgements: The authors thank the Ministry of Health of Djibouti for conducting the 2023 national survey and UNICEF and WFP for their financial support. The authors also acknowledge the University of Senghor (Alexandria, Egypt) for academic support related to doctoral training and for covering the article processing charges upon publication. [page 19, lines 515-518]. |
||
|
|
||
|
4. Response to Comments on the Quality of English Language |
||
|
Comment: The English is fine and does not require any improvement |
||
|
Response: Thank to the reviewer for this appreciation. |
||
Author Response File:
Author Response.pdf
Round 2
Reviewer 2 Report
Comments and Suggestions for Authors Many thanks for sending through this revised version. I consider that the authors have responded very well to the issues that I raised and that the revised MS can be accepted for publication.