Review Reports - HIV Status and Non-Communicable Diseases in Kenya: Sex-Stratified Findings from the Kenya Demographic and Health Survey 2022

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Title: Well-coined

Abstract

Well-structured abstract. It clearly communicates the purpose, methods, and key findings.

You report p=0.017, p=0.003, and p=0.070. The third value (p=0.070) is not statistically significant at the conventional 0.05 threshold. The way it is written alongside significant results could mislead readers. Consider either:

Labelling it explicitly: (p=0.070, not statistically significant) Or restructuring the sentence to separate significant from non-significant findings more clearly.

The phrase "some cancers" at the end is unexpectedly vague after a precise, quantitative abstract. Either name the specific cancers observed or remove this phrase entirely if cancer data were not formally analysed.

Verify the age range 15-54 years is the correct DHS age range used. The standard Kenya DHS age range for women is 15–49 years, and for men 15–54 years. If both sexes are combined in one sentence, clarify whether the range differs by sex or state the broader range with a note.

The abstract does not explicitly state that this is a cross-sectional study using secondary data analysis. Most journals require the study design to be stated clearly, ideally in the first or second sentence.

Introduction is well written

This is a well-constructed introduction that follows the classic funnel structure — moving from global NCD burden → HIV-NCD intersection → SSA/Kenya-specific evidence → research gap.

This is strong and sets the context well. Two issues to address:

"NCD" should be "NCDs" (plural) to be consistent with usage throughout.

The statement "cardiovascular disease being the most frequently documented NCD at 88.9%" (lines 98–99) is striking — an 88.9% CVD prevalence among PLHIV would be extraordinary. Please verify whether this is the prevalence of CVD within those who had NCD comorbidity (i.e., 88.9% of the 11.5%).

Materials and Methods

This section is clearly written, logically organised, and contains the essential elements for a secondary data analysis paper.

There is a numerical inconsistency requiring verification. The total respondents with known HIV status are stated as 23,887 (13,819 women + 10,068 men = 23,887. However, the full survey included 14,818 men and 33,137 women (total 47,955). The subset of 23,887 with known HIV status represents about 49.8% of the total sample - it would strengthen transparency to briefly explain why the remaining ~24,000 were excluded (e.g., HIV status missing or not tested).

The age range inconsistency flagged earlier reappears: women are eligible aged 15–49 (line 132) but the survey population is described as 15–54 years for both sexes (line 130). Clarify this explicitly - the differing age ranges for men and women should be acknowledged as a limitation or at least stated clearly in the population description.

The NCD list "hypertension, diabetes, heart disease, lung disease, arthritis, depression, anxiety, and certain cancers") - the phrase "certain cancers" again appears vague. Name the specific cancer types collected in KDHS 2022, or state "cancer (type not specified)" to be accurate.

The occupation categories (unemployed, agriculture, service-manual, non-manual) would benefit from a brief note on how "unemployed" was defined whether it includes homemakers, students, or those seeking work - as this affects interpretation, particularly given the sex-stratified design.

Results: Described in detail

Remove % symbol from each row. Instead, include it with the n

What proportion of NCDs population had multi-morbidity or clustering of NCDs

Discussion

The Discussion opens well with a concise summary of key findings before moving into comparative literature

The summary accurately mirrors the abstract findings: higher prevalence of tobacco use, alcohol use, depression, hypertension, and cervical cancer among PLHIV; no significant difference in diabetes or breast cancer. This is appropriate. The Discussion should restate key results without reintroducing numbers already in the Results section. However, you do restate some AOR figures here. Choose one approach: either cite them briefly for emphasis (acceptable) or omit them and refer readers to the Results table. Mixing both styles within the paragraph looks inconsistent.

When you cite studies showing higher NCD burden among PLHIV, you should note whether those studies were also conducted in LMIC or sub-Saharan African settings if your citations are from high-income countries on ART-suppressed cohorts, the comparability is limited and should be acknowledged.

The sentence discussing cervical cancer and HIV (immunosuppression → HPV persistence → cervical dysplasia) is scientifically sound but is stated as a single brief clause. Given that cervical cancer is a notable finding and one of the few AIDS-defining cancers that is preventable, this mechanism deserves one full explanatory sentence.

Conclusion

This section has to be written well, incorporating different sustainable strategies to address NCDs among PLHIV rather than stating what is stated in the results section.

Comments for author File: Comments.pdf

Author Response

Reviewer 1

Title: Well-coined

Abstract

Well-structured abstract. It clearly communicates the purpose, methods, and key findings.

Labelling it explicitly: (p=0.070, not statistically significant) Or restructuring the sentence to separate significant from non-significant findings more clearly.

Response: Thank you for these helpful comments. We have added clear methodological phrasing in the opening sentence identifying the study as a cross-sectional, secondary data analysis of the Kenya Demographic and Health Survey.

Separation of Significant vs. Non-Significant P-Values: Differentiated outcomes achieving standard statistical significance () from those that attenuated or were borderline. Explicitly reported depression in women () and anxiety across sexes (men: ; women: ) as statistically significant. Explicitly labelled adjusted depression in men () and adjusted cervical cancer in women () as not statistically significant at the conventional threshold. We have replaced “some cancers” with “cervical cancer,” and clarified the age ranges (15–54 years for men and 15–49 years for women).

Introduction is well written

This is a well-constructed introduction that follows the classic funnel structure — moving from global NCD burden → HIV-NCD intersection → SSA/Kenya-specific evidence → research gap.

This is strong and sets the context well. Two issues to address:

"NCD" should be "NCDs" (plural) to be consistent with usage throughout.

Response: Thank you for these comments. We have reviewed the use of “NCD” throughout the manuscript and now use “NCDs” when referring to multiple diseases, while retaining “NCD” for a single disease or as a modifier, such as in “NCD comorbidity.”

We also agree that the original wording could be read as suggesting that 88.9% of all PLHIV had cardiovascular disease. We have clarified that 11.5% of PLHIV had a recorded NCD, and that cardiovascular disease was the most frequently documented condition within this subgroup (88.9%). Line 101-102 in the clean revised version.

Materials and Methods

This section is clearly written, logically organised, and contains the essential elements for a secondary data analysis paper.

Response: Thank you for these comments. We have clarified that of the 47,955 total survey respondents, HIV serotesting was administered to a nationally representative 50% household subsample per standard KDHS biomarker protocol. Within this subsample, 23,887 individuals (13,819 women and 10,068 men) completed testing with valid results. We now state the sex-specific age ranges: 15–49 years for women and 15–54 years for men. We have also replaced “certain cancers” with “breast and cervical cancer” and clarified how the occupation categories were derived, including who was classified as “unemployed.”

Results: Described in detail

Remove % symbol from each row. Instead, include it with the n

What proportion of NCDs population had multi-morbidity or clustering of NCDs

Response: Thank you. We removed all % symbols from the data rows and the column headers. We formatted the column headers and secondary sub-headers explicitly as (%) [or Unweighted (Weighted %)], ensuring that each cell reports the frequency followed by the percentage in parentheses without redundant symbols. We updated the group comparison headers from "Men/Women in General Population" to "HIV-Negative Men" and "HIV-Negative Women" to accurately describe the survey reference categories and avoid confusion with total population aggregates. We have also included information on the proportion of NCD population with multimorbidity.

Discussion

The Discussion opens well with a concise summary of key findings before moving into comparative literature

Response: Thank you for this suggestion. We have removed the percentages and adjusted odds ratios repeated in the Discussion opening paragraph. The paragraph now summarises the direction of the findings, with the estimates retained in the Results section and tables.

We have expanded the Discussion to explain how HIV-associated immunosuppression may contribute to HPV persistence and progression to cervical cancer, and have noted the implications for prevention and screening.

Conclusion

This section has to be written well, incorporating different sustainable strategies to address NCDs among PLHIV rather than stating what is stated in the results section.

Response: Thank you for this suggestion. We have revised the Conclusion to focus on sustainable strategies rather than restating the Results. It now proposes integrating NCD risk assessment and care into routine HIV services, strengthening referral and follow-up pathways, and improving access to cervical cancer screening and treatment for women living with HIV.

Reviewer 2 Report

Comments and Suggestions for Authors

Dear Authors,

This manuscript addresses an important public health topic by examining differences in non-communicable diseases (NCDs) and related risk factors according to HIV status and sex using data from the 2022 Kenya Demographic and Health Survey. The nationally representative dataset and sex-stratified approach are important strengths. However, several aspects of the methodological reporting, variable definitions, statistical analysis, and interpretation require clarification before the findings can be fully evaluated. I therefore recommend Major Revision. My main comments are as follows.

Definition and measurement of study variables

The definitions of several outcomes and risk factors require clarification and greater consistency. Most importantly, the Methods describe cervical and breast cancer screening, whereas the Results and Table 2 present "Cervical Cancer" and "Breast Cancer" as disease outcomes. Please verify what the underlying KDHS variables actually represent and ensure that screening history is not interpreted as cancer prevalence unless the variables specifically capture diagnosed disease. The terminology should be corrected consistently throughout the manuscript.

The operational definitions of tobacco and alcohol use should also be clarified, as different descriptions are provided across the Methods. In addition, the manuscript should distinguish clearly between objectively measured variables, self-reported diagnoses, behavioral self-reports, and screening-related measures. This is particularly relevant because hypertension is objectively assessed using blood pressure measurements and/or medication use, whereas the Strengths and Limitations section more generally describes the NCD measures as self-reported.

Complex survey design and statistical analysis

The statistical methods require more detailed reporting. Although the manuscript states that analyses were weighted to the Kenyan population, it is unclear whether the full KDHS complex survey design, including sampling weights, clustering, and stratification, was incorporated into descriptive analyses, chi-square tests, confidence intervals, and multivariable regression models. Please clarify exactly how the survey design was handled and identify the relevant statistical procedures used.

For the logistic regression models, please also clarify the dependent variables, HIV reference category, covariates and their coding, and the rationale for adjustment. Given the relatively small HIV-positive subgroups and the rarity of some outcomes, the precision and stability of the corresponding estimates should also be considered when interpreting wide confidence intervals.

Verification and interpretation of the regression results

Table 2 contains an important ambiguity regarding the direction of the odds ratios. The headings indicate "HIV-Negative vs PLHIV," whereas the Results interpret odds ratios greater than 1 as higher odds among PLHIV. Please verify the coding and reference category and ensure that the table, Methods, Results, and Discussion are fully consistent.

The distinction between unadjusted and adjusted findings should also remain explicit, particularly where statistical significance changes after adjustment, as occurs for depression and hypertension among men. Such changes may be entirely plausible after covariate adjustment, but the descriptive and adjusted findings should not be presented interchangeably. I also recommend re-checking all estimates, confidence intervals, and p-values in Table 2 against the statistical output. For example, the reported aORs for depression and breast cancer among women are 3.368 and 3.367, respectively. Their numerical similarity does not necessarily indicate an error, but it provides an additional reason for careful verification of the table.

Analytical sample and Table 1

The analytical sample requires clearer reporting. The Methods state that 23,887 participants had known HIV status, comprising 10,068 men and 13,819 women. However, Table 1 sums correctly for men (130 + 9,938 = 10,068) but not for women, where 332 + 13,357 = 13,689 rather than 13,819. Please reconcile these figures and clearly describe the progression from the total KDHS sample to the final analytical sample, including relevant exclusions or missing observations.

Several percentages in Table 1 should also be verified against the analytical output. In particular, some age-group percentages among participants living with HIV show repeated numerical increments that are difficult to reconcile with the stated group sizes. Please re-check the denominators, survey weighting, and calculation of the percentages and clearly distinguish unweighted sample counts from weighted percentages or estimates. Table 1 also contains blank rows and minor typographical or formatting errors that should be corrected.

Interpretation, limitations, and conclusions

The Discussion should distinguish more consistently between observed associations and possible explanations. Biological, behavioral, and psychosocial mechanisms such as ART-related effects, inflammation, stigma, or coping behavior may reasonably be discussed, but these mechanisms were not directly examined in the present cross-sectional analysis. Similarly, differences in self-reported diagnoses may partly reflect differences in healthcare contact, screening, or likelihood of diagnosis rather than differences in underlying disease prevalence alone.

The Strengths and Limitations section should therefore be strengthened to address the cross-sectional design, residual confounding, limitations of self-reported measures where applicable, the relatively small HIV-positive subgroups, and uncertainty surrounding uncommon outcomes. The statement implying that all NCD measures were self-reported should also be corrected.

The Conclusions are currently very brief and somewhat broader than the individual findings support. They should be modestly expanded to summarize the principal sex-specific findings, distinguish statistically supported associations from descriptive differences, and present recommendations for integrated HIV and NCD care as implications informed by the findings rather than direct effects demonstrated by this study.

Abstract, organization, and reporting consistency

The Abstract should reflect the complete analytical approach. It currently mentions descriptive statistics and chi-square tests but omits the multivariable logistic regression analyses that produced the adjusted odds ratios. The analytical sample should also be reported clearly rather than giving the impression that the entire KDHS sample contributed to the HIV-stratified analyses.

Please perform a final consistency check across the Abstract, Results, tables, and Discussion for sample sizes, prevalence estimates, p-values, adjusted estimates, and rounding. For example, sex-specific tobacco and alcohol prevalence estimates reported in the Discussion should first be presented clearly in the Results or corresponding tables. Figure 1 is cited in the Results but appears to be absent from the submitted manuscript and should be included or the citation corrected.

Finally, the headings within Section 2 should either be formatted consistently as numbered subsections (2.1, 2.2, etc.) or standardized according to journal style. The "Strengths and Limitations" heading should be treated similarly. Please also review the Author Contributions statement to ensure that all initials correspond to listed authors, clarify the ethics/IRB status applicable to this secondary analysis in accordance with journal requirements, and undertake a final English-language and editorial proofread.

Overall, the study addresses a relevant question using a valuable population-based dataset. The principal concerns appear addressable through clarification and verification of the existing analyses, improved methodological reporting, correction of internal inconsistencies, and more proportionate interpretation of the findings.

Comments on the Quality of English Language

The manuscript is generally understandable, but careful language editing is recommended to improve grammar, sentence structure, terminology, and consistency throughout the text.

Author Response

Definition and measurement of study variables

Response: Thank you for identifying this inconsistency. We checked the KDHS questions and the variables used in our analysis. The cancer outcomes were based on respondents reporting that a healthcare worker had told them they had breast or cervical cancer, rather than on screening history. We have corrected the Methods and revised the table headings and manuscript text to describe these as self-reported diagnoses. We have reviewed the Methods to ensure that the definitions of tobacco and alcohol use are consistent throughout and clearly identified both as self-reported behaviours. We have also clarified that hypertension was determined using measured blood pressure and reported antihypertensive medication use. Finally, we revised the Strengths and Limitations section to distinguish these measures from self-reported diagnoses rather than describing all NCD measures as self-reported.

Complex survey design and statistical analysis

Response: We have revised the Methods and Discussion sections to detail our statistical procedures and address estimate precision:

Complex Survey Architecture: We clarified that all analyses utilized Stata’s survey (svy) estimation to account for the KDHS two-stage stratified cluster sampling. This included specifying primary sampling units, stratification, and normalized individual sampling weights. Bivariate comparisons now report design-based Rao-Scott corrected Pearson chi-square tests.
Regression Specifications: Multivariable models were fitted separately for each binary NCD outcome (hypertension, diabetes, depression, anxiety, and cancers). We explicitly designated HIV-negative status as the reference category (0=HIV negative, 1=PLHIV ). Models adjusted for a priori confounders: age group, residence, education, marital status, and wealth quintile.
Estimate Precision and Stability: We added text contextualizing how small HIV-positive subgroups (130 men, 332 women) and rare conditions (e.g., diabetes, site-specific cancers) resulted in sparse data cells and wide confidence intervals, reflecting statistical imprecision.

Verification and interpretation of the regression results

Response: We have corrected the Odds Ratio Orientation: Corrected all headings now labeled as Table 3, Methods, Results, and Discussion from "HIV-Negative vs PLHIV" to "Odds Ratio (PLHIV vs. HIV-Negative)" to align with interpretations of OR representing higher odds among PLHIV.

Explicit Separation of Crude and Adjusted Findings: Disentangled unadjusted from multivariable findings across the manuscript. Explicitly highlighted that while crude odds were statistically significant for hypertension (p=0.039) and depression (p=0.005) in men, both attenuated after sociodemographic adjustment (hypertension: aOR = 1.44, ; depression: aOR = 2.21, ).

Full Audit of Table 3 Estimates: Re-checked all estimates against the primary Stata log files. We have now resolved the draft overlap between depression and breast cancer in women by confirming their verified multivariable values: depression aOR = 2.70 ( p=0.001) and breast cancer aOR = 3.45 (p=0.137).

Side-by-Side Presentation: We have restructured Table 2 to present unadjusted and adjusted odds ratios alongside their respective 95% confidence intervals and -values, enabling direct comparison of the effect of covariate adjustment.

Analytical sample and Table 1

Response: See response to reviewer 1’s comment on material and methods viz: "Of the 47,955 total survey respondents, HIV serotesting was administered to a nationally representative 50% household subsample per standard KDHS biomarker protocol. Within this subsample, 23,887 individuals (13,819 women and 10,068 men) completed testing with valid results"

Interpretation, limitations, and conclusions

Response: Thank you for this suggestion. We have expanded the Strengths and Limitations section to address the cross-sectional design, which precludes conclusions about temporal relationships or causality; the possibility of residual confounding; and recall or reporting bias in the self-reported measures. We have also noted that the smaller HIV-positive subgroups and low numbers of uncommon outcomes may reduce the precision of some estimates. Finally, we corrected the statement that implied all NCD measures were self-reported: blood pressure was measured, while medication use and relevant diagnoses were self-reported.

In the conclusion, we have revised and expanded the section, framing integrated HIV and NCD care as a potential implication of the findings, rather than an effect evaluated or demonstrated by this cross-sectional study.

Abstract, organization, and reporting consistency

Response: Thank you for these helpful comments. We have addressed the issues raised here. Specifically, we revised the Abstract to include the multivariable logistic regression analyses and clearly report the analytical sample; verified consistency across the Abstract, Results, tables, and Discussion; ensured that all prevalence estimates discussed are presented in the Results or tables; corrected the Figure 1 citation; standardized the section headings; verified the Author Contributions statement; clarified the ethics approval applicable to this secondary analysis; and completed a final English-language and editorial review.

Reviewer 3 Report

Comments and Suggestions for Authors

It was a pleasure to review the article entitled “HIV Status and Non-Communicable Diseases in Kenya: Sex-Stratified Findings from the Kenya Demographic and Health Survey 2022“ which describe the relationship between HIV and Non-Communicable Diseases, with people living with HIV having more behavioral risks, mental health issues, and cancer. The article indicating the need and importance of prevention and management strategies for Non-Communicable Diseases and HIV care, with special attention to mental health.

I have some comments that would need to be addressed.

Materials and Methods section

Page no. 3, line no. 121 “A cross-sectional design was used to analyse data from the 2022 Kenya…..” , mention here the period.

Results

Page no. 5, table 1, the age group Men Living with HIV (n=130) is not matching the total of 100%. Please check the total throught the table.

Page no. 6, table 1, Line no. 262, the % sign is double here. Please check this.

Page no. 7, line no. 288, “Men living with HIV had lower odds of diabetes compared to men who were HIV- negative (aOR = 0.64, p = 0.658)” thought, this association was not statistically significant, what could be the possible reason?

Author Response

I have some comments that would need to be addressed.

Materials and Methods section

Page no. 3, line no. 121 “A cross-sectional design was used to analyse data from the 2022 Kenya…..” , mention here the period.

Response: Thank you for this comment. The Methods section already states that the 2022 KDHS data were collected from 17 February to 19 July 2022. We have now clarified in the same sentence that this study was a cross-sectional secondary analysis of those data.

Results

Page no. 5, table 1, the age group Men Living with HIV (n=130) is not matching the total of 100%. Please check the total throught the table.

Response: Table 1 now updated.

Page no. 6, table 1, Line no. 262, the % sign is double here. Please check this.

Response: We have revised this.

Response: Thank you for this observation. Although the point estimate was below 1, the association was not statistically significant (p = 0.658), so we cannot conclude that men living with HIV had lower odds of diabetes. The estimate may reflect chance variation, particularly if few men in the HIV-positive subgroup reported diabetes. We have revised the Results to avoid interpreting the direction of this estimate.

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

Dear Authors,

Thank you for the careful revision and detailed responses. The manuscript has improved substantially, and the principal methodological concerns raised in the previous review have been addressed. Only a few reporting and internal-consistency issues remain.

1. Please ensure that the interpretation is fully aligned with the revised results.
Several statements appear to remain from the previous analysis. The cervical cancer association is described as significant despite aOR = 4.21, p = 0.053 (95% CI 0.98–18.11); hypertension among men is discussed as significant despite aOR = 1.44, p = 0.390; and depression among men is similarly presented as significant despite aOR = 2.21, p = 0.069. The Discussion also refers to an “eight-fold increase” in cervical cancer, which no longer corresponds to the current estimates. Please revise these statements and check the significance markers in Table 3 throughout, as several do not correspond to the reported p-values.

2. Please reconcile the remaining methodological and denominator inconsistencies.
The Abstract and Methods list different adjustment variables (“county” versus “wealth quintile”), and the definitions of tobacco and alcohol use differ between sections of the Methods. The reported female analytical sample (13,819) also does not reconcile with the female counts in Table 1 (13,689), and some Table 1 categories do not sum to their group totals. Please clarify missing data/denominators and whether the percentages are weighted or unweighted.

3. A final consistency check would resolve the remaining reporting issues.
Multimorbidity (18.4%) is reported in the Abstract and Results but is not clearly defined in the Methods, and the cited Table 2 does not appear to contain these data. Please also harmonize the terminology for the mental-health outcomes, explain the analytical denominators for breast and cervical cancer, and ensure that the IRB/ethics information described in the response letter is included in the final manuscript.

Overall, the major concerns raised in the previous review have been satisfactorily addressed. The remaining points are limited to interpretation, reporting, and internal consistency and should be straightforward to correct. Once these are addressed, I have no further substantive concerns.

Author Response

Response 1: Thank you for highlighting these inconsistencies. We have revised the Discussion and Table 3 to ensure that all interpretations, effect estimates, p-values, and significance markers are fully aligned with the revised results.

Response 2: We thank the reviewer for identifying these inconsistencies. We have reconciled each point as follows:

Adjustment Covariates ("County" vs. "Wealth Quintile"): We have harmonized the list of adjustment variables across the Abstract, Methods, and Table 3. The multivariable models controlled for age group, type of place of residence (urban/rural), educational attainment, marital status, and county cluster. Wealth quintile was examined purely as a descriptive sociodemographic characteristic in Table 1 and was not entered into the regression models to prevent multicollinearity with education, residence, and geography.

County Clustering: While KDHS collects data across all 47 counties, stratifying by county alongside sex and HIV status leads to sparse cells and statistical non-convergence for low-prevalence outcomes (e.g., cancers and diabetes). We therefore clustered the 47 counties into eight established regional blocs (Coast, North Eastern, Eastern, Central, Rift Valley, Western, Nyanza, and Nairobi). This grouping reflects established administrative jurisdictions and contemporary regional health and economic cooperation blocs (such as the Lake Region Economic Bloc and Frontier Counties Development Council). Through this, we capture the macro-georgraphic variations in health systems but also in the model used, the degree of freedom is maintained.

Wealth Quintiles: We clarified that wealth quintiles are derived directly from the standard DHS Wealth Index, which uses principal component analysis (PCA) of household durable assets, housing infrastructure, and utility access to rank individuals into five national quintiles (Lowest to Highest).

We have expanded the Methods section to include both detailed descriptions.

Definitions of Tobacco and Alcohol Use: We have reviewed and aligned the operational definitions across all sections of the Methods. The text now consistently defines tobacco use as current use of any tobacco products (binary: non-user vs. user) and alcohol consumption as any current drinking status within the preceding 30 days (binary: non-drinker vs. drinker).
Female Analytical Sample Size: We corrected the female analytical sample size in the Methods, and Results to . The previously reported figure of 13,819 was an inadvertent error resulting from an earlier draft that added the 130 men living with HIV to the female total (). The verified sample sizes are for men ( PLHIV, HIV-negative) and for women ( PLHIV, HIV-negative).
Table 1 Subgroup Totals and Missing Data: We added an explanatory footnote to Table 1 clarifying that variables such as Occupation ( for men; for women) represent restricted survey modules administered only to respondents eligible for employment questions in the KDHS questionnaire, which accounts for the discrepancy with the overall totals. Similarly, ART use was evaluated only among respondents living with HIV ( men; women).
Weighted vs. Unweighted Percentages: We have clarified in the Table 1 footnote that Table 1 presents unweighted counts () and unweighted percentages (%). In contrast, Table 2 and Table 3 report survey-weighted prevalence percentages and survey-weighted multivariable logistic regression models taking into account the complex survey sampling design (sampling weights, clustering, and stratification)

Response 3: We thank the reviewer for pointing out this oversight. We have resolved this across the Methods, Results, and Table 2:

Methods Section: We added an explicit operational definition of multimorbidity under the Measures subsection: "Multimorbidity was defined as the concurrent presence of two or more chronic conditions (among hypertension, diabetes, depressive symptoms, anxiety symptoms, and cancer). In our sub-analysis, we estimated the proportion of adults presenting with two or more concurrent conditions among those with at least one diagnosed chronic condition."
Table 2: We updated Table 2 to include a summary entry and footnote presenting the survey-weighted multimorbidity estimates (18.4% overall; 20.8% in men and 17.2% in women) among individuals with chronic conditions.
Results Section: We refined the narrative in the Results to explicitly link the 18.4% statistic to its denominator and the updated Table 2.