Data Availability Statement

R: There are no ethical or legal restrictions preventing the sharing of the anonymized dataset used in this study. The dataset has been fully anonymized prior to public release and does not contain any direct identifiers or information that could reasonably lead to participant identification. The study was approved by the Research Ethics Committee of the State University of Pará (UEPA), approval number 4.915.684, and all procedures complied with Brazilian ethical regulations and the Declaration of Helsinki.

The minimum dataset required to reproduce the analyses reported in this manuscript, together with the corresponding codebook describing all variables and coding procedures, is publicly available and may be accessed through the following DOI: https://doi.org/10.1371/journal.pntd.0014078.s001

This repository contains the anonymized dataset supporting the findings of the study, including sociodemographic, environmental, territorial, and health-related variables used in the statistical and machine-learning analyses. Therefore, there are no restrictions on data access, and no data access committee is required. All interested researchers may freely access the dataset through the DOI provided above.

Funding Statement

R: This study received no specific funding from any public, commercial, or not-for-profit funding agency. Although some co-authors are undergraduate medical students affiliated with Metropolitan College of Pará – FAMETRO, Santarém, Pará, Brasil and University of Amazonia – UNAMA, Santarém, Pará, Brasil, these institutions did not provide financial support, salaries, research materials, infrastructure, or any other resources for this study. The research was conducted within the framework of the State University of Pará (UEPA), a public university. The participation of students from FAMETRO and UNAMA occurred on a voluntary basis and was unrelated to any institutional sponsorship, employment relationship, or financial support. Therefore, no commercial entity had any role in the study design, data collection, data analysis, decision to publish, or preparation of the manuscript.

Competing Interests Statement

The authors declare that they have no competing interests. The editorial office identified affiliations with Metropolitan College of Pará – FAMETRO, Santarém, Pará, Brasil and University of Amazonia – UNAMA, Santarém, Pará, Brasil as potential commercial affiliations. However, the authors affiliated with these institutions are undergraduate students and are not employees, consultants, shareholders, or representatives of these organizations. These institutions did not provide funding, salaries, materials, infrastructure, or any form of support for this research. The participation of these authors was entirely voluntary and independent of their academic affiliations. This does not alter our adherence to PLOS ONE policies on sharing data and materials.

REVIEWER 1

1. Misalignment between the study objective and study design. The manuscript aims to “identify determinants” of healthcare access. However, given the cross-sectional design, this wording overstates the analytical capacity of the study. Cross-sectional studies do not establish temporality and are not suitable for identifying determinants in a causal sense. The objective should be revised to reflect the exploratory and associative nature of the analysis.

Response: We thank the reviewer for this observation. We clarify that we consider healthcare access as a multidimensional and territorially conditioned phenomenon, a concept derived from the Social Determinants of Health framework widely used in health research, particularly in the Amazon region and strongly linked to the healthcare system adopted in Brazil. To improve methodological precision and avoid causal interpretations, we revised the study objective by replacing the verb “identify” with “investigate.” The objective now reads: “The present study aimed to investigate social and territorial determinants of healthcare access in the Brazilian Amazon.”

2. Conceptual weakness of the primary outcome (“healthcare access”). The outcome is based on self-reported receipt of care when needed, which may conflate need, demand, and access and is subject to recall and reporting biases. A more robust conceptual framework is needed to support this operationalization and avoid overinterpretation.

Response: To provide a more robust conceptual framework regarding the theoretical conception of access adopted in this study, we revised the manuscript and added the following text to the Methods section (lines 151–159): “This approach was adopted considering the organization of health services in the Amazon region and the geographical and organizational characteristics of the communities studied, which are related to the effective provision of care in the context of the healthcare system. In this sense, access no longer depends exclusively on spontaneous demand or individual perception of need, but rather on the organization and interrelation of services, referral and counter-referral processes, the implementation of active search strategies, the presence of a multidisciplinary team in the territory, and the ability to travel to access or continue using the healthcare network.”

3. Insufficient discussion of potential biases. Major sources of bias are not adequately addressed, including selection bias (community-based recruitment) and information bias (self-reported measures).

Response: We thank the reviewer for this observation. We expanded the discussion of methodological biases present in the study, with particular emphasis on information bias. In addition, we detailed the possible implications of recall bias, social desirability bias, and community-based recruitment. The following sections were revised: “Second, several variables were based on self-reported information, which may be subject to recall bias, leading to the oversimplification or underreporting of older events, an overemphasis on more memorable experiences at the expense of routine healthcare visits, and distorted perceptions resulting from the fragmented care context experienced in the region. In addition, they may be influenced by social desirability bias, stemming from embarrassment when reporting information, possible influence by interviewers, and the historical context of vulnerability of the studied population, generating fear of negative consequences for the community or personal relationships with the health service.”

“The study was conducted in quilombola communities within a specific region of the Brazilian Amazon, whose territorial and logistical characteristics (population dispersion, seasonal and precarious displacement, high-cost transportation, and geographic isolation) increase the complexity of fieldwork and hinder population recruitment...”

4. Absence of a descriptive table of sample characteristics. The Results section does not present a comprehensive table describing the study population. This limits interpretability and the assessment of external validity. Inclusion is recommended according to the STROBE Statement.

Response: A table describing the sociodemographic characteristics of the analyzed sample was included.

5. Potential overinterpretation of community-level results. Results based on aggregated data should be interpreted cautiously, considering the limited number of communities and the risk of ecological inference.

Response: We thank the reviewer for this observation. We revised the Discussion and Conclusion sections to provide a more cautious interpretation of the findings at the community level, correcting passages that could suggest extrapolation or generalization beyond the inferential scope of the study. We also reinforced the study limitations to make clear the need for caution when interpreting and generalizing the findings.

REVIEWER 2

1. It is necessary to include a map contextualizing the location of the communities used in the study, to provide a better understanding of how isolated these communities are, since not all readers are familiar with the Amazonian reality.

Response: A map was included to contextualize the analyzed communities.

2. I would recommend changing the title to: “Healthcare Access Inequalities in Quilombola Rural and Riverside Communities in the Brazilian Amazon: Social, Organizational, and Territorial Determinants.” With the original title, it appears that three different community types (quilombola, rural, and riverside) are being analyzed separately, whereas the methodology makes it clear that these characteristics refer to the same communities.

Response: We thank the reviewer for the suggestion. The title was revised and changed to: “Healthcare Access Inequalities in Quilombola Rural and Riverside Communities in the Brazilian Amazon: Social, Organizational, and Territorial Determinants.”

3. Line 125: Remove “either”.

Response: Removed.

4. Lines 174–176: Information repeated from lines 161–164.

Response: Verified and redundant information was removed.

5. Line 193: Lasso and Ridge (only the first letter capitalized).

Response: Verified and corrected.

6. Lines 202–203: Repetition of information from lines 200–201.

Response: Verified and redundant information was removed.

7. Line 205: Replace “when class imbalance was present” with “In cases of class imbalance”.

Response: Verified and replaced.

8. Lines 211–212: Replace with “Variables included in clustering were selected to reflect access-related characteristics; post-access outcomes were excluded, as the goal was to describe access profiles.”

Response: Verified and replaced.

9. Line 215: Replace with “was adapted for the healthcare access outcome”.

Response: Verified and replaced.

10. Line 223: Replace with “lower socioeconomic status”.

Response: Verified and replaced.

11. Line 234: “Among the 517 records in the database, 512 participants...” This information already appears in the Methods section.

Response: Verified and lines 234–235 were removed.

12. Line 277: Table 3 is unnecessary because the information is repeated in the text.

Response: Verified and Table 3 was removed.

13. Line 288: Table 4 is unnecessary because all information is repeated in the text.

Response: We chose to retain the table and revise the preceding text into an interpretative synthesis of the findings, avoiding repetition of coefficients already presented in the table and reducing redundancy.

14. Line 281: Change to “Lasso” and “Ridge”.

Response: Verified and corrected.

15. Line 322: Table 7 is unnecessary because all information is repeated in the text.

Response: We chose to retain the table and removed the sentence “The community-level random-intercept standard deviation was 0.201 (variance = 0.040), corresponding to an approximate ICC of 0.012,” maintaining the interpretation of the findings while avoiding duplication.

16. Line 336: Table 9 is unnecessary because all information is repeated in the text.

Response: Numerical values and the phrase “with additional contributions from education (−0.424), employment status (−0.349), and household income (−0.335)” were removed from the text preceding Table 9, maintaining the interpretation while avoiding repetition.

17. Line 393: Lasso, only the first letter capitalized.

Line 415: Correct to “context”.

Response: Verified and corrected.

18. Line 429: Replace “assumes more dramatic contours” with “becomes more severe”.

Response: Verified and replaced.

19. Question: I would like to know why telehealth services were not discussed in greater depth. There are many articles on this topic, including in the Amazon region itself. It is mentioned only briefly. In my opinion, this topic should be addressed given its relevance.

Response: We thank the reviewer for this observation and recognize the relevance of telehealth as a strategy to expand healthcare access, particularly in the Brazilian Amazon. We added a paragraph addressing this issue. However, we chose not to discuss it in greater depth because telehealth/telemedicine services were not available in the communities investigated during the study period and were not identified by participants as part of their healthcare access experience. Therefore, although the topic is relevant to the broader discussion, it was not part of the investigation or analytical framework of the present study and was mentioned only as a potential facilitator of healthcare access.

Attachments

Attachment

Submitted filename: LETTER OF REPLY.pdf

RESPONSE TO THE EDITOR AND REVIEWERS

Manuscript title: Healthcare Access Inequalities in Quilombola Rural and Riverside Communities in the Brazilian Amazon: Associations with Social, Organizational, and Territorial Factors

Dear Editor and Reviewers,

We sincerely thank the Editor and the Reviewers for the careful re-evaluation of our manuscript and for the detailed and constructive recommendations. We appreciate the recognition that the previous revision improved the methodological description, contextualization, and presentation of the findings.

We agree with the Editor that some of our previous responses described the changes made without sufficiently explaining the scientific rationale underlying those changes. We therefore conducted a comprehensive revision of both the manuscript and the response letter. In the present response, each item specifies: (1) what was changed; (2) where the change can be found; and (3) how the modification addresses the methodological, conceptual, or editorial concern raised.

All changes have been highlighted in the revised manuscript.

ADDITIONAL COMMENTS FROM THE EDITOR

Editor comment 1:

Please ensure that all reviewer comments are addressed in full. In several comments, the response letter states that a change was made, but the revision does not always fully resolve the underlying concern. In particular, the conceptual definition of healthcare access should be strengthened while explicitly acknowledging the limitations of the operational definition, and the limitations of the cross-sectional design should be discussed more explicitly regarding temporality and causality.

Response:

We thank the Editor for this important observation. We agree that our previous response did not fully distinguish the broad theoretical construct of healthcare access from the narrower operational measure used in this study.

Three substantive changes were made.

First, we reformulated the objective so that it no longer suggests causal identification. The revised objective now reads:

“The present study aimed to examine individual, organizational, and territorial factors associated with self-reported healthcare access among quilombola rural and riverside communities in Santarém, Brazilian Amazon, using conventional statistical and machine-learning approaches.”

This wording appears in the Abstract, lines 22–26, and at the end of the Introduction, lines 108–113. The terms “examine” and “associated with” were selected because the cross-sectional design permits the examination of associations and exploratory patterns but does not establish temporal ordering or causal determination.

Second, we substantially revised the operational definition of the primary outcome in the Methods, lines 162–179. Healthcare access is now explicitly defined as self-reported realized access, namely, receipt of healthcare after the participant perceived a need during the previous 12 months. We also state that this measure captures only one stage of the access process and does not represent a comprehensive assessment of healthcare accessibility. The manuscript now explicitly acknowledges that the measure does not fully distinguish health need, perceived need, demand, care-seeking, service availability, utilization, timeliness, continuity, or adequacy of care.

Third, we expanded the limitations in the Discussion, lines 485–527, to explain that the cross-sectional design precludes establishing whether the social, territorial, and organizational characteristics preceded or resulted from the reported access experience. We therefore state that the findings must be interpreted as associations and hypothesis-generating patterns rather than causal effects.

These revisions address the Editor’s concern by aligning the conceptual interpretation of the outcome and the reported findings with the actual measurement strategy and the inferential limits of the study design.

Editor comment 2

The responses should explain why, and not only what. Each response should clearly explain what changes were made, where the changes can be found, and how they address the reviewer’s concern.

Response

We agree and have completely restructured the response letter. Each response now includes: the specific textual, methodological, or analytical modification; the section and current line numbers in which the modification appears; an explanation of why the modification resolves or mitigates the concern raised.

Responses that previously stated only “verified,” “corrected,” “removed,” or “revised” have been replaced with complete explanations. For minor grammatical corrections, we identify the revised passage and explain that the modification improves linguistic accuracy or removes redundancy without altering the substantive meaning. For methodological issues, we explicitly describe the inferential or conceptual implications of the revision.

Editor comment 3

When a reviewer’s recommendation is not followed, provide a scientific justification. The authors are not required to implement every suggestion, but retaining a table or text should be scientifically justified.

Response

We thank the Editor for this clarification. We have expanded the responses concerning Tables 4, 7, and 9, which were retained despite the Reviewer’s recommendations for removal.

Table 4 was retained because it reports the complete coefficients from the Lasso and Ridge models and enables readers to evaluate the direction, relative magnitude, and consistency of the associations across the two penalized approaches. The accompanying Results text was shortened so that it now provides only an interpretative synthesis rather than repeating all coefficients. The revised text and table are located in the Results, lines 326-339.

Table 7 was retained because reporting the community-level random-intercept standard deviation, variance, and approximate ICC is necessary for transparent evaluation of contextual heterogeneity. These parameters cannot be adequately evaluated from a narrative statement alone. Numerical duplication was removed from the preceding paragraph, which now focuses on substantive interpretation. The revised paragraph and table are located in the Results, lines 361-372.

Table 9 was retained because interpretation and reproducibility of a PCA-derived index require transparent reporting of component loadings. The loadings show how each socioeconomic and structural variable contributed to the first component and allow readers to assess the substantive meaning of the Social Vulnerability Index. Numerical values were removed from the narrative and retained only in the table. The revised material appears in the Results, lines 375-383.

These decisions preserve the information necessary for analytical transparency while eliminating textual duplication.

Editor comment 4

Please verify consistency throughout the manuscript. In particular, revisions to the study objective should be reflected consistently in the Abstract, Introduction, manuscript metadata, cover letter, and response letter.

Response

We performed a systematic terminology review across the manuscript and accompanying submission documents.

The study objective has been standardized as:

“To examine individual, organizational, and territorial factors associated with self-reported healthcare access among quilombola rural and riverside communities in Santarém, Brazilian Amazon, using conventional statistical and machine-learning approaches.”

This wording appears in the Abstract, lines 22-26, and the Introduction, lines 108–113, and has also been incorporated into the revised cover letter, submission-system metadata, and response letter.

We also revised the study-design terminology. The study is now described in the Methods, lines 114–120, as a: “community-based cross-sectional survey”, rather than a population-based study. The same wording has been introduced in the Abstract.

Throughout the Results, Discussion, and Conclusion, causal formulations were revised, particularly when referring to the empirical findings. Terms such as “caused,” “determined,” “proved,” and “confirmed” were replaced, where appropriate, by “was associated with,” “showed a pattern,” “suggested,” or “was consistent with.”

The Conclusion, lines 528-536, now explicitly states that the findings represent associations and exploratory patterns and do not establish temporal or causal relationships.

The revised title was also aligned with this interpretation: “Healthcare Access Inequalities in Quilombola Rural and Riverside Communities in the Brazilian Amazon: Associations with Social, Organizational, and Territorial Factors.”

This terminology audit ensures consistency between the study objective, design, outcome definition, analytical strategy, and inferential claims.

Editor comment 5

Please carefully review the revised Data Availability Statement. If the dataset is now publicly available, ensure that the repository corresponds to the dataset used in the present study and that the statement is consistent with the submission system. If availability changed since the original submission, clearly explain the change.

Response

We thank the Editor for identifying the need to clarify the change in the Data Availability Statement. In the original submission, we adopted a conservative position regarding public data sharing because the study involved participants from small and geographically identifiable quilombola communities. We were particularly concerned about the possibility of indirect identification through combinations of territorial, demographic, and socioeconomic variables, even after the removal of direct personal identifiers.

Following the first review round, the dataset was reassessed and a minimum analytical dataset was prepared for public sharing. Direct identifiers were removed, and variables presenting a potential risk of indirect identification were reviewed, recoded, or organized in a manner intended to preserve participant confidentiality. The resulting dataset contains the observations and analytical variables required to support the findings reported in the manuscript, together with a codebook describing variable definitions, categories, and coding procedures.

The Data Availability Statement was revised in the manuscript, lines 575-584, and the same wording has been entered into the journal submission system. The revised statement reads:

“The anonymized dataset supporting the findings of this study is publicly available as Supporting Information. The dataset includes the variables analyzed in this study, comprising sociodemographic, environmental, territorial, and health-related information, as well as the corresponding codebook describing variable definitions, categories, and coding procedures used in the statistical and machine-learning analyses. All data were fully anonymized prior to sharing to ensure participant confidentiality and compliance with ethical standards for research involving vulnerable populations.”

The public file and its DOI were verified against the final analytical dataset before resubmission. Thus, the change from the initial statement reflects the preparation and verification of an anonymized minimum dataset, rather than a change in the ethical approval or an assumption that the original identifiable database could be shared without restriction.

Dataset DOI: https://doi.org/10.1371/journal.pntd.0014078.s001

Editor comment 6

Please ensure that the limitations explicitly discuss information bias associated with self-reported variables, selection bias related to community-based recruitment, and the limited number of communities included in contextual analyses and its implications for generalization.

Response

We agree and have substantially expanded the limitations section in the Discussion, lines 485-527. Regarding information bias, we now explain that healthcare need, receipt of care, and problem resolution were self-reported using a 12-month reference period. These measures may be affected by recall error, variation in individual interpretation, and social-desirability bias. We also state that these processes may have resulted in outcome misclassification and that, if reporting differed according to socioeconomic characteristics, community, previous service experiences, or interviewer interaction, the observed associations could have been attenuated or exaggerated.

Regarding selection bias, we now explicitly describe recruitment as community-based and non-probabilistic. Residents were included according to their presence, availability, and willingness to participate during fieldwork. We explain that this process may have overrepresented residents with greater availability or community engagement and underrepresented workers, mobile residents, people in less accessible areas, and individuals unwilling to participate. We further clarify that the sample-size calculation does not guarantee population representativeness.

Regarding contextual analyses, we now state that only nine communities were included and that their sample sizes were markedly unequal. Some communities contained very few participants, and analyses restricted to communities with at least 10 observations involved an even smaller number of aggregate units. Consequently, ecological correlations, variance components, and ICC estimates may be unstable and sensitive to individual communities. We explicitly state that community-level associations cannot be transferred directly to individual-level relationships and should be regarded as descriptive and hypothesis-generating.

We clarify that the findings may not be generalizable to other quilombola, urban, Indigenous, or Amazonian populations with different health-service networks, mobility conditions, and sociocultural characteristics.

REVIEWER 1

Reviewer 1 - Comment 1

Misalignment between the study objective and study design. The manuscript aims to “identify determinants” of healthcare access. However, given the cross-sectional design, this wording overstates the analytical capacity of the study.

Response

We thank the Reviewer and agree that the previous objective could be interpreted causally.

The objective was reformulated in the Abstract, lines 22–26, and the Introduction, lines 108–113, as follows:

“The present study aimed to examine individual, organizational, and territorial factors associated with self-reported healthcare access among quilombola rural and riverside communities in Santarém, Brazilian Amazon, using conventional statistical and machine-learning approaches.”

We replaced “identify determinants” with “examine factors associated with” because a cross-sectional study can estimate associations and describe patterns but cannot establish the temporal ordering required for causal interpretation.

The title, conclusion, cover letter, submission metadata, and response letter were also revised to use association-based terminology. In addition, the limitations and conclusion now explicitly state that the results do not establish temporal or causal relationships.

Reviewer 1 - Comment 2

Conceptual weakness of the primary outcome. The outcome is based on self-reported receipt of care when needed, which may conflate need, demand, and access and is subject to recall and reporting biases.

Response

We thank the Reviewer and agree that our previous description did not sufficiently distinguish the theoretical construct of healthcare access from its operational measurement in this study.

The primary outcome was redefined in the Methods, lines 162-179, as self-reported realized access: “The primary outcome was healthcare access operationalized as self-reported realized access, defined as having received healthcare when the participant perceived that care was needed during the previous 12 months.”

We now explicitly state that this measure captures only the successful receipt of care following a perceived need. It should not be interpreted as a comprehensive measure of healthcare accessibility and does not distinguish all stages between objective health need, perceived need, care-seeking, service availability, utilization, continuity, or adequacy.

We also explain that territorial mobility, transportation, primary-care organization, referral pathways, and health-system responsiveness were examined as contextual

Attachments

Attachment

Submitted filename: Response to Reviewers_L.docx