Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

General Evaluation

In this manuscript, the authors evaluate associations between pet-keeping characteristics and perceived stress among older adults aged across five provinces in China. Moving beyond a binary "owner vs. non-owner" comparison to explore granular caregiving factors—such as direct caregiving frequency, pet illness, and walking habits—is an important, timely, and clinically relevant objective in human-animal interaction research.

However, substantial conceptual, methodological, and analytical vulnerabilities in the study require the authors to address them prior to publication. Specifically, the paper co-opts the "One Health" paradigm without evaluating environmental, veterinary, or biological endpoints; overlooks clustering effects inherent to multicenter sampling; may fail to operationalize key exposure metrics or measure pet headcount; and attempts to draw directional, causal conclusions from a cross-sectional snapshot.

Below, I provide detailed critiques and constructive suggestions to assist the authors in strengthening their manuscript.

Major Comments

1. One Welfare rather than One Health

        The authors frame this study under a "One Health perspective" across the Title, Abstract, Introduction, Discussion, and Conclusion. I must emphasize that this framing may be conceptually inaccurate. The formal One Health framework fundamentally requires an integrative evaluation across human, animal, and environmental interfaces. This study includes no environmental or green space determinants, no zoonotic or biological transmission data, no clinical veterinary records, and no animal physiological or welfare measures. The data rely strictly on human self-reported questionnaires administered at human community health service centers. The dynamics examined—daily caregiving chores, financial/time burdens, pet illness as an emotional stressor, and human-perceived stress—are classic issues of Human-Animal Interaction (HAI), caregiver burden, or at most One Welfare (the link between animal well-being and human social/emotional health).

       I suggest that the authors remove the phrase "One Health" replacing it with accurate terminology such as "One Welfare," "Caregiver Burden," or "Human-Animal Interaction". Re-anchor the theoretical framework in the Introduction and Discussion around the One Welfare paradigm and literature on companion animal caregiving strain.

2 Title 

     I find the current title problematic for two primary reasons:   1. The phrase 'from a One Health Perspective' is scientifically inaccurate. As detailed above, the study does not evaluate environmental components, zoonotic risks, or veterinary diagnostic data. Retaining 'One Health' in the title misrepresents the methodology and theoretical boundaries of the paper.   2. The title refers generically to 'pet-related characteristics,' yet the paper's core finding is that perceived stress reduction was driven strictly by exclusive dog ownership, with null findings for cats and dual-pet households.   I suggest the authors revise the title to accurately describe the scope of the study—for example: 'Associations between Companion Animal Caregiving Characteristics and Perceived Stress among Chinese Older Adults: A Multicenter Cross-Sectional Study' (or frame it under a 'One Welfare' perspective if animal welfare and caregiver burden are appropriately discussed)."

3. Limitation Of samples

      The narrative repeatedly asserts that pet ownership "reduces stress," "buffers stress," or provides "protective" benefits. However, in a cross-sectional study, it is scientifically impossible to state whether participants became better or worse after acquiring a pet. Without pre-adoption baseline measurements, there is no way to evaluate intra-individual change.

• Confounding by indication (Self-Selection)? An older adult who was already experiencing severe loneliness, depression, or distress may have actively acquired a pet as a self-directed coping strategy. Even if the animal provided substantial therapeutic relief, their post-adoption stress score might still register as higher than or equal to that of a well-supported non-owner who never needed a pet to improve their life.
• Healthy Adopter Bias? Conversely, the modest score differences observed (33.96 vs. 35.27) may simply reflect self-selection of healthier, wealthier, and more resilient seniors into pet ownership.

I suggest that the authors explicitly state in the Discussion and Limitations that only prospective, longitudinal studies with pre- and post-acquisition assessments can definitively determine whether companion animals improve or worsen older adults' psychological trajectories. Eliminate all directional and evaluative phrasing implying improvement, alleviation, or causal protection from the Abstract, main text, and Conclusion, restricting statements strictly to cross-sectional associations.

4. Missing Definitions & Pet Count

        Section 2.2.2 leaves essential variables open to arbitrary interpretation. The authors never define what constitutes "direct pet care" (e.g., does it encompass feeding, grooming, cleaning waste/litter, administering medication, or tactile affection?). The response categories ("rarely," "occasionally," "frequently") lack objective behavioral time anchors (such as minutes/day or days/week). Furthermore, "pet illness frequency" lacks both a defined recall window (e.g., past 30 days vs. pet lifetime) and veterinary diagnostic criteria, conflating mild transient issues with severe, chronic disorders. The survey failed to distinguish whether the older participant was the primary guardian or merely co-residing with pets owned and managed by adult children in multi-generational households.

       More Pets, More Problems? In lines 258–261, the authors hypothesize that care burdens in multi-pet homes offset emotional benefits. However, the survey never recorded the actual number of pets kept per household. A household with one toy dog is pooled identically with a household managing four large dogs or five cats. Caring for multiple animals exponentially multiplies physical chores, financial outlays, inter-pet behavioral conflict, and the risk that at least one animal is chronically ill (which had the highest adverse effect size in the study, =+4.33). By omitting quantitative pet count, the authors introduced severe exposure misclassification.

        I suggest that the authors provide the exact wording and operational definitions for all pet-related survey items in Section 2.2.2. Explicitly acknowledge the absence of quantitative pet headcount data as a major design limitation in the Discussion.

5. Superficial Explanation of Cat Ownership Findings

         In Table 3, exclusive cat ownership showed a significant association with lower stress in unadjusted models, but completely collapsed to non-significance after full multivariable adjustment (=-0.98,p=0.17). The authors dismiss this finding in just two sentences (lines 254–256) by stating it is "consistent with US-based findings". They provide no behavioral, domestic, psychological, or cultural explanation for why cats failed to buffer perceived stress in this cohort.

         If cat ownership does not demonstrate measurable stress reduction in cross-sectional models, why is cat guardianship surging so rapidly among urban older adults? The literature on feline guardianship reveals that the primary motivation for acquiring a cat is often seeking an emotional "healer" or comforter—meaning prospective owners frequently enter guardianship with pre-existing emotional vulnerabilities, psychological distress, loneliness, or empty-nest syndrome.

         Because cross-sectional data lack pre-adoption baselines, this creates confounding by indication: cat owners may be using feline companionship as self-care/self-medication to cope with higher baseline distress. While the cat may provide comfort and subjective happiness during difficult emotional episodes, the owner's net perceived stress score remains elevated or comparable to non-owners who never experienced such emotional distress in the first place.

         In addition, cat owners frequently keep multiple cats rather than a single animal. While multiple cats provide emotional enrichment, they compound indoor maintenance, financial expenditures, and veterinary demands. Periods of peace and comfort alternate with acute caregiving strain, creating a complex emotional dynamic that a static sum-score like the PSS-14 fails to capture when animal headcount is not controlled for.

I suggest that the authors expand the Discussion to address:

• Discuss how the intent to adopt a cat often stems from pre-existing psychological strain or loneliness. Acknowledge that cats may serve an active coping role for vulnerable individuals, explaining why cross-sectional scores appear non-significant when compared against well-adjusted non-owners who did not require an animal to improve their life.
• Note that feline guardianship commonly involves multiple animals. Without recording cat headcount per home, the authors could not isolate the emotional benefits of single-cat companionship from the compounding logistical and financial friction of multi-cat households.
• Address the realities of urban Chinese domestic environments, where cats are kept strictly indoors. Daily litter box scooping, odor control, floor sanitation, and indoor shedding represent unremitting domestic chores that offer no outdoor physical exercise or neighborhood social facilitation.
• Contrast the communicative ecology of cats with dogs. Dogs demand structured, outward daily routines (leashed walks, overt greeting rituals, obedience) that enforce behavioral activation. Cats exhibit high behavioral autonomy and subtle communicative cues, which may not translate into the same structured psychological stress-buffering in older adults.

6. Statistical problems

• Participants were sampled across 3 to 4 community health centers in five distinct provinces. Applying standard Ordinary Least Squares (OLS) multiple linear regression ignores the nested, hierarchical data structure. Observations within the same clinic or province share unobserved regional, economic, and healthcare accessibility factors. This violation of independence likely underestimates standard errors and inflates Type I error rates.
• Table 2 and Table 3 use Model 1, 2, and 3, but Table 4 and Table B1 switch to "Univariate" and "Multivariate" analysis. The authors must clarify whether the multivariate models in Table 4 and Table B1 adjust for all Model 3 socioeconomic/health covariates (income, chronic disease, education, etc.), or only mutually adjust for the four pet-care indicators. If demographic covariates were omitted, the findings in Table 4 suffer from massive omitted-variable bias.
• In Table B1, daily dog walking flips from being significantly protective in univariate analysis (=-2.50,p<0.05) to positive and non-significant in multivariate analysis (=+0.69,p=0.56). Furthermore, ownership duration >3 years is strongly protective across all pets (Table 4), but completely loses significance among dog owners alone (Table B1: =-1.27,p=0.22). The authors never address these parameter instabilities or suppression effects.

       I suggest that the authors to re-analyze the data using Multilevel Mixed-Effects Linear Modeling (LMM) or Generalized Estimating Equations (GEE) with cluster-robust standard errors to account for center- and province-level clustering. Clearly list all adjusted covariates in the footnotes of Table 4 and Table B1. Report collinearity diagnostics (Variance Inflation Factors / VIF) to verify whether direct care and walking frequency introduce multicollinearity. Fully detail the multiple imputation methodology in Section 2.2.1 (missing data percentages per variable, imputation algorithm, and number of imputed datasets ).

7. Dog Walking vs. Caregiver Burden

       The authors repeatedly assert that pet walking was "obligation-driven" to justify why walking was linked to higher stress (lines 22, 42, 234, 266–269, 325). This conclusion is deeply flawed due to the ample literature that confirms that routine dog walking enhances canine welfare, reduces behavioral issues, and provides bidirectional cardiovascular and mental health benefits to owners. Labeling routine walking as an inherent "burden" contradicts established human-animal interaction evidence.   

       The survey assessed "pet walking" across all pet owners. Walking is species-inappropriate for most domestic cats, causing acute distress for the cat and frustration for the owner. Conflating canine and feline walking under one generic variable introduces a severe exposure error.   

       The survey only recorded walking frequency, never measuring motivation or subjective obligation. Real caregiver burden in older adults is dynamic and driven by pet illness, multi-pet management, owner physical frailty/mobility issues, and financial strain—not the act of walking a dog itself.   I suggest that the authors remove all claims that pet walking is inherently "obligatory" or burdensome. Instead, they should reframe their discussion around the true drivers of burden (health costs, multiple pets, physical limitations) and properly differentiate species-specific needs, rather than pooling dogs and cats into a generic walking measure. 

8. Limitations & Future Directions

The limitations section is presented as a single, dense paragraph (lines 298–317) that only covers generic cross-sectional caveats. Furthermore, the manuscript lacks a dedicated, actionable "Future Directions" section.

I suggest that the authors restructure the Limitations into distinct thematic sub-paragraphs covering: (a) cross-sectional design and lack of pre-post data (confounding by indication); (b) lack of household pet headcount data; (c) unvalidated single-item exposures lacking clinical veterinary verification; (d) selection bias from clinic-based sampling; and (e) unmeasured functional independence metrics (e.g., ADL/IADL). Add a clear Future Directions section outlining the necessity of prospective pre-post adoption cohorts, objective veterinary assessments, and qualitative investigations into caregiving motivations.

9. Absence of Veterinary or Animal Behavior Expertise

All authors are based in human General Practice and Medical departments, with no co-authors or collaborators from veterinary medicine or animal welfare/behavior. This interdisciplinary gap directly weakens the study's validity:

• "Pet illness" is measured only as a crude, subjective impression ("rarely, occasionally, frequently") without clinical staging, chronicity, or diagnostic criteria.
• Findings such as walking-related stress and the lack of cat-ownership benefits are explained through speculative psychosocial narratives, while core ethological realities (e.g., leash reactivity, poor training, inter-cat resource guarding, indoor confinement stress) are ignored.
• A true One Welfare or One Health investigation cannot be conducted solely through a human medical lens.

I strongly suggest that the authors collaborate with veterinary or companion animal behavior colleagues to re-examine their measures and interpret these human-animal interactions with proper clinical and ethological rigor.

Author Response

1. One Welfare rather than One Health

Comment: The “One Health” framing is conceptually inaccurate because the study includes no environmental, zoonotic, veterinary, or biological endpoints; the dynamics examined are classic issues of HAI, caregiver burden, or at most One Welfare. The authors should remove “One Health” and replace it with accurate terminology and re-anchor the framework.

Response: We fully agree with this assessment. We have replaced “One Health” with “One Welfare” throughout the title, abstract, keywords, introduction, methods, discussion, and conclusions. The only remaining occurrences of the phrase “One Health” appear (i) within the titles of cited references, which must be preserved verbatim, and (ii) in one sentence that explicitly contrasts the traditional One Health focus on zoonotic risks with the One Welfare focus of this study.

The Introduction now defines the One Welfare framework, in which animal welfare, human well-being, and the shared domestic environment are interconnected through bidirectional human–animal interactions, citing García Pinillos et al. (newly added reference [9]) and Tarazona et al. (newly added reference [10]). The Discussion now explicitly links companion-animal welfare to the burdens and rewards of caregiving, and the term human–animal interaction is used where appropriate.

2. Title

Comment: The title is problematic because “from a One Health Perspective” is scientifically inaccurate, and the generic phrase “pet-related characteristics” does not reflect the core finding that the association is driven by exclusive dog ownership with null findings for cats and dual-pet households.

Response: Thank you for your advice. We have revised the title to: “Associations between Companion Animal Caregiving Characteristics and Perceived Stress among Chinese Older Adults: A Multicenter Cross-Sectional Study”

3. Limitation of Samples: Causal Language, Confounding by Indication, and Healthy-Adopter Bias

Comment: The narrative repeatedly asserts that pet ownership “reduces” or “buffers” stress or provides “protective” benefits, which is impossible to establish from cross-sectional data. Confounding by indication and healthy-adopter bias must be addressed, and only prospective longitudinal studies can determine whether companion animals improve or worsen psychological trajectories.

Response: We have removed all directional and evaluative phrasing—including “reduce,” “buffer,” “protective,” and “stress-reducing effect”—from the abstract, results, discussion, and conclusions, and replaced it with strictly associational wording such as “was associated with,” “inverse association,” and “weaker observed associations.”

We now explicitly discuss both mechanisms: confounding by indication, whereby older adults already experiencing loneliness, distress, or empty-nest strain may acquire an animal as a self-directed coping strategy, and healthy-adopter bias, whereby healthier, wealthier, and more resilient seniors may be more able to keep animals. These points appear in the Discussion and in the first point of the Limitations, which now states that only prospective longitudinal studies with pre- and post-acquisition assessments can establish whether companion animals improve or worsen psychological trajectories.

4. Missing Definitions and Pet Count

Comment: Section 2.2.2 does not define “direct pet care,” lacks objective time anchors, gives no recall window or veterinary diagnostic criteria for pet illness, does not distinguish primary guardians from co-residing household members, and never records the number of pets, producing severe exposure misclassification.

Response: Section 2.2.2 now provides full operational definitions. Direct pet care is defined as the hands-on activities of feeding, grooming and bathing, cleaning waste or litter, administering medication, and direct affectionate contact such as petting or holding, reported on a three-level ordinal frequency scale. Pet illness frequency is defined as owner-reported illness episodes, including conditions requiring extra care or veterinary attention, with an explicit statement that no fixed recall window or veterinary diagnostic criteria were applied and that the item therefore reflects subjective perception rather than clinically verified disease.

We also state explicitly that the interview did not distinguish the primary guardian from participants co-residing with animals kept by other family members, and that the number of animals per household was not recorded. The absence of pet headcount, the lack of objective time anchors and clinical verification, and the absence of a primary-guardian distinction are now discussed together as measurement error in the second point of the Limitations.

5. Superficial Explanation of Cat Ownership Findings

Comment: The null adjusted finding for exclusive cat ownership is dismissed in two sentences. The Discussion should address emotional “healer” motivations and confounding by indication, multi-cat households, the realities of indoor keeping in urban China, and the different communicative ecology of cats and dogs.

Response: We have expanded the brief treatment into a dedicated paragraph explaining four mechanisms. First, the decision to acquire a cat often follows pre-existing emotional strain, loneliness, or empty-nest distress, so cats may serve as an active coping or self-comforting strategy for more vulnerable individuals; this confounding by indication can leave post-adoption scores comparable to those of well-adjusted non-owners even when the cat provides genuine comfort (reference [39]).

Second, feline guardianship frequently involves more than one animal, and because cat headcount was not recorded, the benefits of single-cat companionship could not be separated from the compounding housekeeping, financial, and veterinary demands of multi-cat households (references [39,40]). Third, in urban Chinese households cats are typically kept strictly indoors, so daily litter-box cleaning, odor control, floor sanitation, and shedding management constitute recurrent indoor chores that provide no outdoor exercise or neighborhood social facilitation.

Fourth, cats and dogs differ in their communicative ecology: dogs require structured, outward daily routines, including leashed walks and overt greeting rituals, that enforce behavioral activation, whereas cats show greater behavioral autonomy and more subtle communicative cues, which may not produce the same structured psychological activation. We explicitly state that these interpretations remain speculative and warrant dedicated investigation.

6. Statistical Problems: Clustering, Model Specification, Parameter Instability, VIF, and Imputation

Comment: OLS regression ignores the nested multicenter structure and may underestimate standard errors; the adjustment scheme of Table 4 and Table B1 must be clarified; the sign flip of daily dog walking and the attenuation of ownership duration in Table B1 are unexplained; LMM or GEE with cluster-robust standard errors, VIF diagnostics, and full multiple imputation details are requested.

Response: Clustering. Because identifiers for the individual community health service centers were not recorded, clustering was implemented at the province level: the LMM included a province-level random intercept, and the GEE used province as the clustering unit with an exchangeable working correlation. Center-level multilevel models could therefore not be fitted; this is acknowledged in the Limitations. Point estimates were broadly consistent in direction with the primary analyses, although magnitudes were attenuated (as expected for within-province estimates).

Multicollinearity. Variance inflation factors were computed for all predictors; all VIF values were below 3 (maximum 2.57), well below the conventional threshold of 10, indicating no substantial multicollinearity.

Model specification. Footnotes have been added to Table 4 and Table B1 stating that the multivariate models were adjusted for the full covariate set specified in Model 3, namely age, sex, BMI, household type, educational level, marital status, living arrangement, employment status, household monthly income, chronic disease status, smoking status, and drinking status.

Multiple imputation. Section 2.3 now states that BMI was the only covariate with missing data (33 cases, 2.31%, which was also the maximum per-variable missing proportion); these values were imputed using multiple imputation by chained equations (fully conditional specification), generating five imputed datasets, with estimates pooled according to Rubin's rules.

Parameter instabilities. With pet-walking frequency recorded in three categories, daily dog walking was negatively but non-significantly associated with perceived stress in both the univariate (β = −2.50, 95% CI: −5.59 to 0.59) and fully adjusted models (β = −0.82, 95% CI: −2.39 to 0.76), whereas weekly dog walking was positively and significantly associated with higher stress after adjustment (β = 3.83, 95% CI: 2.09 to 5.57). In the small dog-only subgroup (N = 339) the confidence intervals were wide, so these estimates are imprecise rather than evidence of a true reversal; this pattern likely reflects the frequency-only, single-item measurement and the limited subgroup power. Likewise, the attenuation of ownership duration in Table B1 reflects the reduced power of the subgroup and its correlation with owner age and household characteristics; its confidence interval remains compatible with a meaningful association. These estimates should be interpreted together with the cluster-robust results and the measurement limitations.limitation

7. Dog Walking versus Caregiver Burden

Comment: Labeling routine walking as inherently “obligation-driven” contradicts evidence that dog walking benefits canine welfare and owner health; the generic variable conflates dogs with cats for which walking is species-inappropriate; and motivation was never measured, so the true burden drivers (illness, multiple pets, physical frailty, financial strain) should be discussed instead.

Response: We have removed every claim that pet walking is inherently obligatory or burdensome and deleted all uses of “obligation/obligatory” in our own interpretation. We now state that, because the survey measured only walking frequency and did not assess motivation, enjoyment, or perceived obligation, we cannot determine whether participants experienced walking as leisure or as a strain; the observed positive associations may instead reflect underlying burden-related factors, including pet illness, multi-pet management, owners’ physical limitations, and financial strain (new reference [43]).

We acknowledge the evidence that routine dog walking enhances canine welfare and provides bidirectional physical and mental-health benefits, and we now note explicitly that the generic variable pooled dogs and cats, that walking is species-inappropriate for most cats and may cause distress for both animal and owner, and that this introduces measurement error. We recommend that future studies measure motivation, enjoyment, context, and duration in addition to frequency, and we have added the accessibility, safety, and suitability of walking spaces and local dog-access regulations as further contextual factors.

8. Limitations and Future Directions

Comment: The limitations are a single dense paragraph covering only generic caveats, and no dedicated Future Directions section is provided. The Limitations should be restructured into thematic points, and actionable future directions should be added.

Response: The Limitations have been condensed into three concise points presented in a single paragraph: (1) the cross-sectional design and the absence of pre- and post-acquisition data, including confounding by indication and healthy-adopter bias; (2) measurement error, combining the unrecorded pet headcount and resulting exposure misclassification with the single, self-reported exposure items, the lack of objective time anchors, fixed recall windows and veterinary diagnostic criteria, and the absence of a primary-guardian distinction; and (3) generalizability and model specification, combining selection bias from clinic-based recruitment, unmeasured functional independence (ADL/IADL) and cognitive status with residual confounding, and center- and province-level clustering.

The future-research recommendations are integrated into the Discussion and Limitations: prospective longitudinal designs with pre- and post-adoption assessments, objective recording of pet headcount and caregiving time, clinical veterinary assessments and validated animal welfare measures, qualitative investigation of caregiving motivation, obligation, and reward, population-representative sampling, and interdisciplinary collaboration with veterinary and animal behavior colleagues and community-based pet support services.

9. Absence of Veterinary or Animal Behavior Expertise

Comment: All authors are based in human general practice and medical departments, with no veterinary or animal welfare/behavior collaborators, which weakens the validity of the ethological interpretations; collaboration with such colleagues is strongly recommended.

Response: We acknowledge this limitation and do not claim veterinary or ethological expertise. To address it, we have grounded the revised interpretation in veterinary and animal-welfare evidence, including the One Welfare framework (references [9,10]), caregiver burden research (reference [40]), and work on resource negotiation in multispecies households (reference [43]), and we explicitly label the ethological interpretations as speculative. The future-research recommendations commit to interdisciplinary collaboration with veterinary and animal behavior colleagues and with community-based pet support services, and we would welcome the opportunity to involve such colleagues in subsequent work.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

Companionship or Burden? Associations between Pet-Related Characteristics and Perceived Stress among Chinese Older Adults: A Multicenter Cross-Sectional Study from a One Health Perspective

**please see PDF of paper with very minor additional comments and suggested amended/expansions/clarifications**

Overall comments

Thank you for the opportunity to review this interesting and relevant work exploring senior pet caregivers and their perceived stress levels.

This is a subject of increasing relevance with an ageing population and societal changes meaning that companion animals often are a source of comfort, companionship and wider support mechanisms.

The work is clear and has a logical approach and flow. There is ethical oversight and the design and approach are detailed, with the survey being available for view in the document as an appendix. I consider there to be a sensible approach to date collection and methods of analysis (some modelling review is outwith my deep expertise however, so additional reviewer comments with perhaps more expertise in this area is worthy of review). I would like a little more demographic clarity in the methods but this can be largely surmised from the results.

Presentation and analysis of date appears relevant and comprehensive with clear approach and accessibility maintained. Some points are highlighted on the supporting PDF as perhaps worthy of more detailed consideration in the discussion.

The writing, flow and structure is clear, and references are suitable and appropriate. There are a few points noted on the supporting PDF where I feel added evidentiary support for statements might be useful.

The overall work is solid, competent, well considered and does relay some interesting outcomes. I feel the sample is potentially liable to bias (and this is noted but could perhaps have more consideration of the potential wider ramifications of this).

I have added some comments on the supporting PDF for consideration. Data analysis appears relevant and correct, and interpretations are fair and not liable to being overstated, with clear evidence of critical reflection of limitations of the work and study sample.

I really like this work, and feel it has scope to further develop, and expand the content and approach more generally in our understanding of senior pet caregivers and their attitudes and connections to their pets.

References

I have not exhaustively gone through these, but all appear fine in terms of use and suitability. I have not extensively proofread or cross referenced to check validity of use throughout.

Some areas where I feel refs need added are highlighted on supporting PDF as previously noted.

Comments for author File: Comments.pdf

Author Response

1. Additional Demographic Clarity in the Methods

Comment: A little more demographic clarity is requested in the methods.

Response: We have expanded the Methods to describe the mixed quota-and-alternating sampling strategy, the classification of the five provinces into eastern, central, and western regions, the random selection of three to four community health service centers per province, the 1:1 quota-based enrollment protocol, the full eligibility and exclusion criteria, and the complete list of covariates. The operational definitions of all exposure items are now provided in Section 2.2.2, and the baseline distributions are reported in Table 1.

2. Potential Bias of the Sample and Its Wider Ramifications

Comment: The sample is potentially liable to bias, and the potential wider ramifications deserve more consideration.

Response: We have expanded the treatment of selection bias. The third point of the Limitations now explains that recruitment through community health service centers may produce a sample of health-seeking attendees whose health status and behavior differ from those of the wider community-dwelling older population, which constrains the generalizability of the findings. This is discussed alongside healthy-adopter bias and the non-probabilistic quota design, and the future-research recommendations call for population-representative sampling.

3. Additional Evidentiary Support and References

Comment: Some statements would benefit from added evidentiary support and additional references.

Response: We have added a substantial set of references during the revision, including the One Welfare framework [9], the relationship between One Health, One Welfare, and One Biology [10], owners’ relationship perceptions of cats [39], caregiver burden and veterinary service use [40], the bonded green-exercise framework [41], resource negotiation in multispecies households [43], the comprehensive review of dog-walking correlates [44], the co-construction of the human–dog dyadic relationship [45], and the financial cost of pet care [46]. The full reference list has been renumbered in strict order of first citation in the text, so these references are interspersed at their correct positions rather than appended at the end. They support the key new statements in the revised Introduction, Methods, and Discussion.

4. Line-Level Annotations on the Supporting PDF

We have now received the annotated supporting PDF and have responded to every one of the 20 annotations below. For each annotation, the highlighted text and the reviewer’s note are given, followed by our response. The single ink annotation (no. 10) contains no text and is noted for completeness.

Annotation 1

Highlighted text: “60 years and older” (Abstract)

Reviewer note: Please provide the exact age range.

Response: Eligibility was restricted to community-dwelling adults aged 60 years and older, with no upper age limit imposed by the design. The age distribution of the analyzed sample is summarized in Table 1 (mean age 70.93 ± 6.98 years). We have kept the wording consistent with the eligibility criterion and the absence of a pre-specified upper bound.

Annotation 2

Highlighted text: “60 years or older” (Introduction)

Reviewer note: Please provide the exact age range.

Response: This is the same eligibility criterion as Annotation 1. The inclusion threshold was age ≥ 60 years with no upper age cap; the observed sample mean age was 70.93 ± 6.98 years (Table 1).

Annotation 3

Highlighted text: “stress better captures real-world psychological strain” (Introduction)

Reviewer note: Is there evidence for this? A citation is suggested.

Response: The characterization of perceived stress as a measure of everyday, real-world psychological strain is grounded in the PSS literature and the stress-health reviews already cited, namely the PSS validation work [4,18,19,20] and the reviews of stress processes and stress–disease links [2,6]. These sources describe the PSS as capturing general appraisals of unpredictability and overload rather than discrete life events, supporting the statement.

Annotation 4

Highlighted text: “companion-animal-based approaches for stress reduction” (end of Introduction)

Reviewer note: Please also address the support for the companion animals’ own welfare/well-being, in line with One Health–One Welfare.

Response: We agree. The study is now explicitly framed under One Welfare, and the Introduction’s second objective states that we interpret the associations within a framework accounting for both human mental health and companion-animal welfare. The Discussion and Conclusions repeatedly state that interventions must consider both older adults’ psychological well-being and the welfare of the animals, including how frequent illness, indoor confinement, and species-inappropriate walking affect the animals themselves.

Annotation 5

Highlighted text: “60 and above” (Methods)

Reviewer note: Is there an upper age limit?

Response: No upper age limit was set. Eligibility required age ≥ 60 years, sufficient cognitive capacity to consent and complete the assessments, and the exclusion of severe psychiatric disorders, advanced dementia, and terminal illness. The mean age of the sample was 70.93 ± 6.98 years.

Annotation 6

Highlighted text: “Missing values were handled via multiple imputation” (Section 2.2.1)

Reviewer note: The wording is unclear; please add clarification for readers.

Response: We have expanded the explanation. Section 2.3 now describes the fully conditional specification (FCS) multiple-imputation procedure, the number of imputed datasets, and the pooling of estimates according to Rubin’s rules. The final missing-value proportions and number of imputations are marked as values to be inserted.

Annotation 7

Highlighted text: “Pet walking frequency, N(%)” (Table)

Reviewer note: How are cat-only households handled?

Response: The walking item was a generic, frequency-based variable applied across all owners, so cat-only households were included in the same item even though walking is uncommon and often species-inappropriate for cats. The Discussion now states explicitly that the generic variable pooled dogs and cats, that walking is species-inappropriate for most cats and may cause distress to animal and owner, and that this introduces measurement error.

Annotation 8

Highlighted text: “Exclusive cat ownership … lost significance after adjustment” (Results)

Reviewer note: What are the related factors? These will be expanded in the discussion.

Response: We have expanded the interpretation into a dedicated discussion paragraph covering four mechanisms: confounding by indication/emotional “healer” motivations [39], multi-cat households [39,40], strictly indoor keeping and its indoor chores, and the different communicative ecology of cats and dogs. These are explicitly presented as speculative and warranting dedicated study.

Annotation 9

Highlighted text: Weekly/daily dog-walking β coefficients (Results)

Reviewer note: Is this pet walking or dog walking only? The distinction matters, as walking is rare for cats and mainly a dog activity.

Response: The reported variable was a generic “pet walking” frequency item that pooled species; we did not collect a separate dog-only walking measure in the full sample. We have clarified this in the Discussion and report the dog-only subgroup (N = 339) analyses in Table B1, which isolate the dog-walking estimates. The pooling of dogs and cats is acknowledged as a measurement limitation.

Annotation 10

Highlighted text: Ink marking (no text)

Response: This is a hand-drawn ink mark without any accompanying text, so there is no specific comment to act on; we have noted it for completeness.

Annotation 11

Highlighted text: Dog-only subgroup walking β = 3.77 (Results)

Reviewer note: Could this relate to local challenges such as public-space access and dog behavior? Worth considering.

Response: This is a helpful point. We have added a sentence to the walking discussion noting that the accessibility, safety, and suitability of public walking spaces, together with local regulations and restrictions on dog access, may shape whether walking is experienced as leisure or as an obligation. Dog behavior and the practicalities of public-space access are therefore now recognized as contextual factors, and the discussion of burden-related factors (illness, multi-pet management, physical limitations, financial strain) is retained [43].

Annotation 12

Highlighted text: “walking is experienced as a mandatory caregiving duty … obligation” (Discussion)

Reviewer note: Are there cultural differences and is the environment suitable for older adults to walk dogs safely and enjoyably? Dog walking is often part of dog-keeping life rather than a pure obligation, and may relate to how dogs are kept and their freedom.

Response: We agree and have removed any claim that walking is inherently obligatory. The Discussion now states that, because only frequency was measured, we cannot determine whether walking was experienced as leisure or strain, and contrasts recreational walking in Western contexts [42] with the possibility of obligation in our setting. We acknowledge that dog walking is often an integral, enjoyable part of dog keeping and that its experience depends on the living environment, the dog’s freedom and keeping style, and cultural context; the newly added environmental sentence addresses safety and suitability.

Annotation 13

Highlighted text: “future studies … motivation, enjoyment, context, duration” (Discussion)

Reviewer note: Please add the accessibility, safety, suitability, and desirability of dog-walking spaces, as well as local/national regulatory restrictions.

Response: We have incorporated these factors. In addition to motivation, enjoyment, context, and duration, the walking discussion now explicitly cites the accessibility, safety, and suitability of public walking spaces and local regulations/restrictions on dog access as environmental factors that future work should examine.

Annotation 14

Highlighted text: “bonds are bidirectional and evolve” (Discussion)

Reviewer note: This should be linked to the animal-welfare Five Domains model, where interaction with others is already considered part of overall welfare.

Response: We agree that this is consistent with the Five Domains model, in which interactions with other animals and people, alongside the physical domains, contribute to the animal’s overall mental state and welfare. The One Welfare framing throughout the manuscript already treats human–animal interaction and the animal’s welfare as interdependent, and the bidirectional, evolving nature of the bond [45] is discussed in these terms. We have described the concept in the One Welfare language rather than introducing a separate numbered reference.

Annotation 15

Highlighted text: “effects depend on relationship quality and caregiving context” (Discussion)

Reviewer note: This is consistent with other studies, including how pet behavior can negatively affect caregiver health.

Response: We agree. The Discussion already emphasizes that outcomes depend on relationship quality and caregiving context rather than ownership alone, and details how frequent illness and care-related burden can raise stress [40,46]. We recognize that problematic pet behavior, like illness, can add caregiver strain, and this is consistent with the broader caregiver-burden evidence [40].

Annotation 16

Highlighted text: “contact” (Discussion)

Reviewer note: Does this mean physical contact or interaction more broadly? Passive and active interaction can produce similar physiological benefits.

Response: As defined in Section 2.2.2, “direct affectionate contact” includes hands-on physical contact such as petting or holding, while the broader human–animal interaction also covers shared presence and engagement. We acknowledge that both active and more passive forms of interaction can be associated with similar physiological benefits, and the operational definition makes the intended meaning explicit.

Annotation 17

Highlighted text: “Dogs often form stronger attachment bonds” (Discussion)

Reviewer note: This statement needs a supporting citation.

Response: We have added a citation to the attachment-theory literature [31,32], which establishes pets as safe havens and secure bases and frames the stronger attachment-based associations observed for dogs. The citation now appears directly after the statement in the revised manuscript.

Annotation 18

Highlighted text: Limitations: selection bias (Limitations)

Reviewer note: This is an important limitation; how can future research mitigate it?

Response: The future-research recommendations call for moving beyond recruitment through community health service centers to population-representative, probability-based sampling of community-dwelling older adults, which would reduce the selection bias arising from health-seeking attendees. This is stated in the third point of the Limitations and in the Discussion’s future-research sentences.

Annotation 19

Highlighted text: Limitations: self-report/recall (Limitations)

Reviewer note: This is a recognized challenge; please also acknowledge the Hawthorne effect.

Response: The second point of the Limitations already acknowledges recall and social desirability bias from the single, self-reported items. The Hawthorne (observer) effect—participants altering responses or behavior because they are being studied—is closely related to this social desirability/observer bias and is therefore covered by the same acknowledgment; we have clarified this interpretation here.

Annotation 20

Highlighted text: Limitations: residual confounding (Limitations)

Reviewer note: This is an important limitation; please keep it in view.

Response: Residual confounding is explicitly retained in the third point of the Limitations, noting that unmeasured functional independence (ADL/IADL) and cognitive status, alongside other unmeasured variables, may jointly influence caregiving capacity and perceived stress. The cluster-robust sensitivity analyses (LMM/GEE) further address the model-specification concern.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

General Remarks

I want to thank the authors for their thoughtful, rigorous, and detailed revisions. They clearly took the initial feedback to heart. The conceptual shift from "One Health" to One Welfare makes the paper much stronger and far more accurate, and reframing the narrative in terms of caregiver burden, human–animal interaction (HAI), and correlational associations rather than direct causation is a major improvement. The updated analyses—especially addressing multicenter clustering with cluster-robust standard errors, running VIF diagnostics, and laying out the multiple-imputation details—give the empirical findings much firmer grounding.

The paper is very close to publication-ready. However, there are a couple of conceptual points in the Discussion and Limitations that still feel incomplete, along with a few minor editorial oversights that should be cleaned up before final acceptance.

Major Points 

1. State the authors' clinical background in the Limitations

In their response letter (Point 9), the authors candidly noted that their team comes entirely from human General Practice and medicine, with no formal veterinary or animal-behavior collaborators. I appreciate that honesty, and the revised manuscript does a much better job citing relevant veterinary and HAI literature.

However, this point belongs in the actual manuscript. Because this study appears in Veterinary Sciences and discusses pet welfare, caregiving burdens, and pet illness, the authors should be transparent with readers:

• Add a clear sentence in the Limitations section explicitly stating that this study was designed and conducted from a human primary-care medical perspective, without direct veterinary clinical examinations, disease staging, or formal ethological assessments.
• Note that future work at the human–animal interface will benefit substantially from interdisciplinary co-design between general practitioners, veterinary clinicians, and animal welfare scientists.

2. Deepen the interpretation of weekly dog walking

The finding in Table B1 that weekly dog walking is tied to significantly higher perceived stress whereas daily walking is not, remains one of the most interesting results in the paper. While the authors added owner physical limitations as a possible explanation, I believe they missed two very realistic, culturally grounded factors:

• Pets left behind by adult children (involuntary caregiving): In Section 2.2.2, the authors note that they did not distinguish primary owners from older adults simply living with family pets. In urban China, it is common for adult children to buy a dog, move away for work or marriage, and leave the day-to-day care with their aging parents. Walking a high-energy dog that you did not choose to adopt in the first place is a distinct kind of physical and emotional strain.
• Infrequent walking as a symptom of frailty or behavior issues: Daily walking usually reflects a healthy, manageable routine for an active senior with a well-behaved dog. Walking only weekly, on the other hand, is often a red flag—it suggests the owner may have mobility issues, joint pain, or cardiovascular limitations, or that the dog pulls on the leash, reacts to traffic, or acts out around other animals, turning every outing into an ordeal that the owner avoids until absolutely necessary.

I suggest the authors briefly weave these practical realities into lines 312–330. It will make the Discussion feel much more grounded in real-world geriatric and animal-care dynamics.

3. Break up the Limitations section for readability

The authors did a thorough job incorporating all the requested caveats (confounding by indication, healthy-adopter bias, unrecorded pet headcount, single-item self-reports, clinic-based selection, and cluster limitations). However, packing all of this into one massive paragraph (lines 345–365) makes it hard for readers to digest. I suggest splitting it into two or three thematic paragraphs—for example:

• Study design and causality (cross-sectional limits, lack of pre-post data, confounding by indication, healthy-adopter bias);
• Exposure measurement (lack of pet headcount, unvalidated single items, absence of primary-caregiver distinction);
• Sampling and clinical scope (clinic-based sampling, unmeasured functional status/ADLs, and the human medical perspective noted above).

Minor Corrections

1. Table B1 Labeling: Table 4 uses the term "Frequency of direct pet care," but Table B1 labels the same variable "Pet care responsibility." Please standardize this across both tables.

Author Response

1. Clinical Background of the Research Team

Comment: In the response letter (Point 9) the authors candidly noted that their team comes entirely from human General Practice and medicine, with no formal veterinary or animal-behavior collaborators. This point belongs in the manuscript itself: the Limitations should state that the study was designed and conducted from a human primary-care medical perspective, without direct veterinary clinical examinations, disease staging, or formal ethological assessments, and should note that future work at the human–animal interface will benefit from interdisciplinary co-design between general practitioners, veterinary clinicians, and animal welfare scientists.

Response: We agree that this should be transparent to readers rather than confined to the response letter. The third point of the Limitations now states that the study was designed and conducted from a human primary-care medical perspective, without direct veterinary clinical examination, disease staging, or formal ethological assessment, and that the team included no veterinary or animal-behavior collaborators. The future-research paragraph now adds that interdisciplinary collaboration between general practitioners, veterinary clinicians, and animal welfare scientists would substantially strengthen future work at the human–animal interface.

2. Interpretation of Weekly Dog Walking

Comment: The finding that weekly dog walking is associated with significantly higher perceived stress whereas daily walking is not remains one of the most interesting results. Two realistic, culturally grounded factors were missing: pets left behind by adult children (involuntary caregiving), and infrequent walking as a symptom of frailty or animal behavior problems. These practical realities should be briefly woven into the walking discussion (lines 312–330).

Response: We have added these two factors to the walking discussion in brief form, so as not to disrupt the existing argument. First, we note that in urban China dogs are often acquired by adult children who later move away for work or marriage, leaving day-to-day care, including walking, to aging parents—a form of involuntary caregiving that carries its own strain. Second, we note that walking only weekly may be a marker rather than a cause: it can reflect mobility limitation, or a dog that pulls on the leash, reacts to traffic, or is difficult around other animals, turning each outing into an ordeal. The existing discussion of measurement limitations and of the accessibility and safety of walking spaces is retained.

3. Structure of the Limitations

Comment: The Limitations incorporate all requested caveats but are packed into one massive paragraph (lines 345–365), which is hard to digest. They should be split into two or three thematic paragraphs—for example, study design and causality; exposure measurement; and sampling and clinical scope.

Response: We have restructured the Limitations into three thematic paragraphs. The first covers study design and causality: the cross-sectional design, the absence of pre- and post-acquisition data, confounding by indication, and healthy-adopter bias. The second covers exposure measurement: the unrecorded number of animals per household and the resulting exposure misclassification, the single self-reported items without objective time anchors, fixed recall windows, or clinical verification, and the absence of a primary-guardian distinction. The third covers sampling and clinical scope: selection bias from recruitment through community health service centers, unmeasured functional independence (ADL/IADL) and cognitive status, the limited number of province-level clusters and the inability to fit center-level multilevel models, and the human primary-care perspective noted above.

4. Table B1 Variable Label

Comment: Table 4 uses the term “Frequency of direct pet care,” but Table B1 labels the same variable “Pet care responsibility.” This should be standardized across both tables.

Response: We have standardized the label. Table B1 now uses “Frequency of direct pet care,” matching Table 4 and the operational definition given in Section 2.2.2.

Author Response File: Author Response.docx