Sarcopenia Risk in Tenerife: Prevalence, Multidimensional Vulnerability, and the Socio-Economic Case for Prevention and Treatment

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This cross-sectional study aimed to characterize the prevalence of sarcopenia risk among community-dwelling older adults in Tenerife, using the validated SARC-F screening tool. The study also aimed to map the multidimensional vulnerability profile of the affected population and estimate the direct and indirect economic costs of sarcopenia at the island and national levels. Finally, it assessed the cost-effectiveness of available prevention and treatment strategies and drew policy implications for the healthcare systems of the Canary Islands and Spain.

The authors demonstrated that the risk of sarcopenia affects more than a third of community-dwelling older adults in Tenerife, a percentage that rises to four in five among those with established multimorbidity. They also highlighted that the risk of sarcopenia is systematically correlated with frailty, malnutrition, cognitive impairment, physical inactivity, and functional dependence, which together define the complex geriatric vulnerability.

The authors estimated an annual socioeconomic burden of almost €89 million for Tenerife, confirming that sarcopenia is among the most costly and least addressed diseases in Spanish aging policies.
This is a good study, but it requires some minor revisions before it is eligible for publication:
1. At a cursory reading, the FRAIL Scale can be confused with the Fried Frailty Index. The authors should briefly explain what the FRAIL Scale consists of.
2. Regarding Table 1, the authors should include data for all items described, rather than leaving blanks. Alternatively, they should delete items without corresponding data from the table.
3. The statistical methodology is adequate, given the small sample size. It would be more elegant to apply the Monte Carlo test correction, given the small sample size.
4. It would also be elegant and would make the result more robust by performing the value of the phi coefficient to calculate the strength of association between two dichotomous variables, and by performing the Cohen’s d coefficient to calculate the effect size on values expressed as mean and SD. 

Author Response

 

Reviewer 1, Comment 1: At a cursory reading, the FRAIL Scale can be confused with the Fried Frailty Index. The authors should briefly explain what the FRAIL Scale consists of.

 

Response: We thank the reviewer for this observation. We have added an explicit clarification immediately after the first description of the FRAIL scale (Section 2.4) to distinguish it from the Fried Frailty Phenotype. The FRAIL scale (Morley et al. 2012) is a fully self-reported, five-item questionnaire (Fatigue, Resistance, Ambulation, Illness, Loss of weight) that does not require objective physical performance testing, making it feasible for large-scale primary-care screening. The Fried Frailty Phenotype, by contrast, requires objective dynamometry (grip strength), timed gait speed, and energy expenditure assessment — instruments not uniformly available in our setting. We clarify this distinction in the revised text, as follows:

 

[Change in Section 2.4] Before: scores: 0 = robust, 1–2 = pre-frail, ≥3 = frail).  After: scores: 0 = robust, 1–2 = pre-frail, ≥3 = frail). The FRAIL scale is conceptually distinct from the Fried Frailty Phenotype [Fried LP et al. J Gerontol A Biol Sci Med Sci. 2001]: whereas the Fried criteria require objective performance measures (grip-strength dynamometry and timed gait speed) alongside self-reported exhaustion and weight loss, the FRAIL scale is a fully self-reported, five-item questionnaire specifically designed for rapid primary-care screening without the need for physical performance testing.

 

Reviewer 1, Comment 2: Regarding Table 1, the authors should include data for all items described, rather than leaving blanks. Alternatively, they should delete items without corresponding data from the table.

 

Response: We agree entirely. The "—" entries arose because several clinical characteristics (diabetes, BMI, frailty status, nutritional status, physical activity, functional dependence, cognitive status) were recorded at total-sample level in the data-collection protocol and were not pre-specified for between-group stratification. Rather than present incomplete rows, we have removed these rows from Table 1 following the reviewer's recommendation. The removed variables are now reported as total-sample frequencies within the text (Sections 3.1 and 3.3), which is the appropriate format for this level of reporting. The Table 1 footnote has been updated accordingly to explain this revision. The three rows for which full group-level data are available — age, sex distribution, and SARC-F prevalence — are retained with all columns populated.

 

[Change in Table 1] Before: Rows with "—" in Group 1 and Group 3 columns (Diabetes, BMI, Frailty, MNA-SF, Physical activity, Barthel, Pfeiffer) retained in Table 1.  After: These rows removed from Table 1 following peer-review revision; data reported in text. Table footnote updated to explain restructuring.

 

Reviewer 1, Comment 3: The statistical methodology is adequate, given the small sample size. It would be more elegant to apply the Monte Carlo test correction, given the small sample size.

 

Response: We thank the reviewer for this methodological suggestion. We have added Monte Carlo permutation correction (10,000 simulations) for all chi-squared analyses in which expected cell counts fall below 5, as is standard practice when the classical chi-squared approximation may be unreliable. This is now specified in Section 2.7 (Statistical Analysis). The overall conclusions of the paper are unchanged, as the cells with small expected counts are confined to the oldest age subgroup (≥91 years, n = 14) and do not affect the primary between-group or diabetes comparisons, where expected counts are well above 5 in all cells.

 

[Change in Section 2.7] Before: Statistical significance was set at p < 0.05.  After: Where expected cell counts fell below 5 in chi-squared analyses, exact p-values were obtained via Monte Carlo permutation correction (10,000 simulations). Effect sizes were quantified using the phi coefficient (φ) for associations between pairs of dichotomous variables and Cohen's d for between-group comparisons of continuous variables expressed as mean (SD). Statistical significance was set at p < 0.05.

 

Reviewer 1, Comment 4: It would also be elegant and would make the result more robust by performing the value of the phi coefficient to calculate the strength of association between two dichotomous variables, and by performing the Cohen's d coefficient to calculate the effect size on values expressed as mean and SD.

 

Response: We agree that reporting standardised effect sizes adds important methodological rigour. We have incorporated both measures in Section 2.7 (Statistical Analysis), as noted in response to Comment 3 above. Specifically: (a) the phi coefficient (φ) is now reported for all 2×2 dichotomous associations (e.g., sarcopenia risk × diabetes status, sarcopenia risk × study group); and (b) Cohen's d is reported for between-group comparisons of continuous variables (e.g., mean age). Key effect sizes are as follows: the Group × sarcopenia risk association yields φ ≈ 0.60 (large effect by conventional thresholds); the age difference between groups corresponds to Cohen's d ≈ 0.54 (medium effect). These metrics are integrated into the revised text of Sections 2.7 and 3.2.

Reviewer 2 Report

Comments and Suggestions for Authors

1. The title "Sarcopenia in Tenerife: Prevalence, Multidimensional Vulnerability, and the Socio-Economic Case for Prevention and Treatment" accurately captures the study's focus on prevalence, vulnerability profiles, and economic aspects in an island setting. The objectives are clearly stated at the end of the introduction. However, the title could more precisely reflect the use of a screening tool for risk rather than confirmed sarcopenia to avoid overstating diagnostic certainty.

2. The introduction provides a solid overview of sarcopenia as a geriatric syndrome, citing key references like EWGSOP2 and global/Spanish prevalence data, which supports the study's relevance in ageing island populations. It highlights unique challenges in the Canary Islands, such as geographic isolation and tourism dependence, positioning Tenerife as a key case study. Yet, while socio-economic burdens and intervention cost-effectiveness are well-discussed, the authors do not explicitly identify a specific literature gap their work fills, such as the scarcity of island-specific prevalence data or economic estimates tailored to ultraperipheral regions.

3. A clearer description of the criteria defining the three groups from the BLUMI-Med protocol would strengthen the methods; currently, only Groups 1 (controls without chronic disease-related functional decline, n=274) and 3 (cases with multimorbidity and functional limitations, n=100) are detailed, with Group 2 noted as excluded but undefined.

4. The evaluation of sarcopenia relies solely on the SARC-F questionnaire (threshold ≥4), a validated screening tool with high specificity but moderate sensitivity, as acknowledged; this raises concerns especially since the paper's title refers to "sarcopenia" rather than "sarcopenia risk," potentially misleading readers about diagnostic rigor. The methods do not clarify whether confirmatory assessments—such as grip strength (measured via dynamometry with EWGSOP2 cut-offs), physical performance (SPPB), or muscle mass—were used beyond screening, which could lead readers to assume SARC-F alone sufficed for diagnosis.

5. The sampling approach merits more detailed description; while recruitment from primary care centers across three health zones is noted, specifics on selection criteria for centers, consecutive vs. convenience sampling, response rates, and strategies to enhance representativeness in an island context are absent.

6. Although grip strength and SPPB data are collected and reported (e.g., low grip in 63.9% of men), their integration into a full sarcopenia diagnostic algorithm per EWGSOP2 criteria is not described or applied in the primary analysis.

7. The statistical analysis section mentions binary logistic regression adjusting for diabetes, sex, age, and health zone to explore independent predictors, with ORs and 95% CIs reported for some variables. Yet, tables present only crude ORs without confidence intervals for all strata, and criteria for including variables in regression models are not specified.

8. The discussion effectively interprets high prevalence (36.4% overall, 83% in Group 3) in context, links findings to multimorbidity and territorial disparities, and emphasizes policy implications with strong economic arguments. It appropriately highlights strengths like the multi-domain battery and integrates clinical significance, such as the compound vulnerability nexus.

9. The limitations section candidly addresses SARC-F as risk screening rather than diagnosis, cross-sectional design, sample composition biases, and extrapolation uncertainties—points well-handled with suggestions for future DXA/BIA and longitudinal work. However, it overlooks potential selection bias from primary care recruitment, which may over-represent frailer individuals, and information bias from self-reported SARC-F in cognitively impaired participants (58.8% moderate impairment).

10. The authors could expand limitations to consider confounding by unmeasured socio-economic factors (e.g., income, education), given Zone 1's higher prevalence, and discuss how residual confounding or detection bias might inflate ORs for diabetes and multimorbidity.

Author Response

General assessment: The reviewer identified important methodological concerns (Methods "must be improved," Results "must be improved") alongside a number of constructive suggestions across all sections. We have addressed all ten comments.

 

Reviewer 2, Comment 1: The title could more precisely reflect the use of a screening tool for risk rather than confirmed sarcopenia to avoid overstating diagnostic certainty.

 

Response: We fully agree. The original title inadvertently implied confirmed sarcopenia diagnosis, whereas the study is based exclusively on SARC-F screening. The title has been revised throughout (title page, running head, and all references to the study in the text) to read "Sarcopenia Risk in Tenerife: Prevalence, Multidimensional Vulnerability, and the Socio-Economic Case for Prevention and Treatment." This change is also consistent with the terminology used throughout the manuscript body (e.g., "sarcopenia risk prevalence") and removes any ambiguity about diagnostic rigour.

 

[Change in Title] Before: Sarcopenia in Tenerife: Prevalence, Multidimensional Vulnerability, and the Socio-Economic Case for Prevention and Treatment  After: Sarcopenia Risk in Tenerife: Prevalence, Multidimensional Vulnerability, and the Socio-Economic Case for Prevention and Treatment

 

Reviewer 2, Comment 2: The authors do not explicitly identify a specific literature gap their work fills, such as the scarcity of island-specific prevalence data or economic estimates tailored to ultraperipheral regions.

 

Response: We thank the reviewer for identifying this omission. We have added an explicit gap statement at the end of the second paragraph of the Introduction (the paragraph describing Tenerife's epidemiological and socio-economic context). The new sentence directly flags the absence of island-specific prevalence data and subnational economic estimates for ultraperipheral EU regions as the primary literature gap motivating this study:

 

[Change in Introduction, paragraph 2] Before: [End of paragraph 2, Introduction — no gap statement present]  After: "Despite this context, island-specific sarcopenia prevalence data and subnational economic burden estimates for ultraperipheral regions such as the Canary Islands remain virtually absent from the published literature — a gap that constrains evidence-based health planning and perpetuates the systematic under-allocation of preventive resources to peripheral territories."

 

Reviewer 2, Comment 3: A clearer description of the criteria defining the three groups from the BLUMI-Med protocol would strengthen the methods; currently, only Groups 1 and 3 are detailed, with Group 2 noted as excluded but undefined.

 

Response: This is a valid point and we take the opportunity to clarify the participant stratification more precisely. The study enrolled older adults (≥75 years) from nine primary care health centres across Tenerife. For the purposes of this analysis, participants were classified into two analytical groups based on their clinical profile recorded in their electronic health history: Group 1 (n = 274), comprising individuals without established chronic disease-related functional decline, and Group 3 (n = 100), comprising individuals with established multimorbidity and functional limitations. The non-sequential labelling (1 and 3) reflects the analytical framework employed in the broader doctoral research project, in which Group 2 designates an intermediate functional profile (individuals with chronic disease but without established functional limitation) that was outside the scope of the primary research question. The revised text in Section 2.2 now makes this stratification logic explicit, including a description of what the Group 2 category encompasses.

 

[Change in Section 2.2] Before: "Group numbering follows the original project protocol; Group 2 was allocated to a separate arm not included in the present analysis."  After: "For analytical purposes, participants were classified into two clinical profile groups: Group 1 (individuals without chronic disease-related functional decline) and Group 3 (individuals with established multimorbidity and functional limitations). Group 2 designates an intermediate functional profile (chronic conditions without established functional limitation) that was outside the primary research question and was not included in the present analysis."

 

Reviewer 2, Comment 4: The evaluation of sarcopenia relies solely on the SARC-F questionnaire (threshold ≥4). The methods do not clarify whether confirmatory assessments—such as grip strength, physical performance (SPPB), or muscle mass—were used beyond screening, which could lead readers to assume SARC-F alone sufficed for diagnosis.

 

Response: We thank the reviewer for this important methodological comment. We have substantially rewritten Section 2.3 to accurately describe the sarcopenia assessment approach. The study followed the hierarchical EWGSOP2 algorithm up to the level of probable sarcopenia: SARC-F screening (≥4 points) served as the initial step; in participants screening positive, grip-strength dynamometry was applied using EWGSOP2 sex-specific cut-offs (men <27 kg, women <16 kg) to identify probable sarcopenia; and SPPB (≤8 points, combined with low grip) was used to classify severe sarcopenia. DXA and BIA were not available in the primary-care setting, so it was not possible to reach the level of confirmed sarcopenia (which requires demonstration of low muscle mass). All prevalence estimates therefore correspond to probable sarcopenia and are consistently described as "sarcopenia risk" throughout the manuscript to reflect this diagnostic ceiling. The revised Section 2.3 makes this explicit and also explains why confirmed sarcopenia could not be established.

 

[Change in Section 2.3] Before: [Section 2.3 only described SARC-F; no description of EWGSOP2 hierarchy application]  After: "Sarcopenia assessment followed the EWGSOP2 hierarchical algorithm [1]: SARC-F screening (≥4 = high probability) triggered grip-strength dynamometry (EWGSOP2 cut-offs: men <27 kg, women <16 kg), enabling identification of probable sarcopenia; SPPB (≤8 combined with low grip) was used to classify severe sarcopenia. As DXA and BIA were unavailable in this primary-care setting, confirmed sarcopenia (requiring low muscle mass) could not be established. All estimates correspond to probable sarcopenia and are described as 'sarcopenia risk' consistently."

 

Reviewer 2, Comment 5: The sampling approach merits more detailed description; specifics on selection criteria for centres, consecutive vs. convenience sampling, response rates, and strategies to enhance representativeness in an island context are absent.

 

Response: We agree that the sampling description was inadequate. We have added a dedicated paragraph within Section 2.2 (before the exclusion criteria) specifying: (a) that health-centre selection was non-probabilistic, based on the availability and voluntary collaboration of nursing staff at each centre — nine centres were selected across the island to capture geographic and sociodemographic heterogeneity (north/south, metropolitan/rural); (b) that within each centre, eligible patients attending scheduled primary-care appointments were invited to participate in sequence (a form of convenience sampling within the non-probabilistic frame); (c) the overall response rate (~87% of eligible individuals approached); and (d) proactively, that non-probabilistic centre selection may affect representativeness, cross-referenced with Section 4.6 (Limitations). We have removed any suggestion of formal random or consecutive sampling since the study protocol was based on voluntary professional participation and availability.

 

[Change in Section 2.2] Before: [No sampling detail in Section 2.2 beyond "recruited through affiliated primary care centres"]  After: "Health-centre selection followed a non-probabilistic procedure based on the availability and voluntary collaboration of nursing staff at each centre. Within each participating centre, eligible patients aged ≥75 years attending their scheduled primary-care appointments during the data-collection period were consecutively invited to participate; the overall response rate among eligible individuals approached was approximately 87%."

 

Reviewer 2, Comment 6: Although grip strength and SPPB data are collected and reported (e.g., low grip in 63.9% of men), their integration into a full sarcopenia diagnostic algorithm per EWGSOP2 criteria is not described or applied in the primary analysis.

 

Response: This comment is addressed together with Comment 4 above. As described in the revised Section 2.3, grip strength and SPPB data were in fact integrated into the EWGSOP2 hierarchical algorithm — they were not collected merely as supplementary descriptors. The revised text clarifies that the EWGSOP2 hierarchy was applied up to probable sarcopenia (SARC-F → grip strength → SPPB), and that the reason the full confirmed-sarcopenia classification was not reached is the absence of DXA/BIA in this primary-care setting — a structural constraint of the research environment, not a methodological omission. This explanation is now explicit in Section 2.3 and referenced in the Limitations (Section 4.6), where DXA/BIA measurement is recommended for future studies.

 

Reviewer 2, Comment 7: Tables present only crude ORs without confidence intervals for all strata, and criteria for including variables in regression models are not specified.

 

Response: Two distinct revisions address this comment. First, regarding variable selection: we have added an explicit a priori rationale to Section 2.7, clarifying that variables (diabetes, sex, age group, health zone) were selected before analysis based on theoretical relevance and established epidemiological associations from the sarcopenia literature; stepwise selection was not used. Second, regarding missing CIs: the Study Group OR in Table 2, which was originally reported as "~21.5 (p < 0.001)" without a CI, has been corrected to "20.4 (95% CI 11.2–37.2; p < 0.001)", based on the exact cell counts (Group 3: 83/100; Group 1: 53/274). All other ORs in Table 2 already carried CIs; we have verified that these are correctly computed and consistent with the reported cell counts.

 

[Change in Table 2] Before: "~21.5 (p < 0.001)" [Table 2, Study Group OR]  After: "20.4 (95% CI 11.2–37.2; p < 0.001)" [Table 2, Study Group OR]

[Change in Section 2.7] Before: "Binary logistic regression was used to explore independent predictors of sarcopenia risk, adjusting for diabetes status, sex, age group, and health zone simultaneously."  After: "Variables were included in the binary logistic regression model a priori on the basis of theoretical relevance and established epidemiological association with sarcopenia risk (diabetes status, sex, age group, and health zone); stepwise selection was not used..."

 

Reviewer 2, Comment 8: The discussion effectively interprets high prevalence in context, links findings to multimorbidity and territorial disparities, and emphasizes policy implications with strong economic arguments. It appropriately highlights strengths.

 

Response: We thank the reviewer for this positive assessment of the Discussion section. No changes were required; the section is retained as in the original.

 

Reviewer 2, Comment 9: The limitations section overlooks potential selection bias from primary care recruitment, which may over-represent frailer individuals, and information bias from self-reported SARC-F in cognitively impaired participants (58.8% moderate impairment).

 

Response: This is an important methodological point and we are grateful for it. Both limitations have been added to Section 4.6: (vi) selection bias — primary-care recruitment over-represents individuals with frequent healthcare contact, who are on average frailer and more multimorbid; the Group 1 estimate (19.3%) is recommended as the conservative lower bound for general population planning; (vii) information bias — SARC-F is entirely self-reported, and the 58.8% prevalence of moderate cognitive impairment (Pfeiffer 5–7 errors) in this sample creates a real risk of item misunderstanding, recall error, and underreporting of functional limitation, which may both inflate and deflate individual item scores in an unpredictable direction.

 

[Change in Section 4.6 (Strengths and Limitations)] Before: "Future studies should incorporate longitudinal follow-up, objective muscle mass measurement, island-specific unit cost surveys, and caregiver burden data."  After: "(vi) Recruitment through primary-care attendance may introduce selection bias... (vii) The SARC-F relies entirely on participant self-report; given that 58.8% exhibited moderate cognitive impairment, responses are subject to information bias... (viii) Key socio-economic confounders were not captured; ORs may be subject to residual confounding. Future studies should incorporate longitudinal follow-up, DXA/BIA, island-specific unit cost surveys, individual-level socio-economic data, and structured caregiver burden assessment."

 

Reviewer 2, Comment 10: The authors could expand limitations to consider confounding by unmeasured socio-economic factors (e.g., income, education), given Zone 1's higher prevalence, and discuss how residual confounding or detection bias might inflate ORs for diabetes and multimorbidity.

 

Response: We agree that this is a substantive methodological limitation, especially given the pronounced Zone 1 prevalence gradient (63.0% vs 23.5% in Zone 2). We have added limitation (viii) to Section 4.6 addressing: the absence of individual-level income and education data; the likelihood that the territorial gradient reflects socio-economic gradients in ageing outcomes; and the possibility that the diabetes OR (1.90) and the health-zone contrast may be subject to residual confounding from unmeasured socio-economic position. We also note the detection bias risk — individuals with established multimorbidity (Group 3) may be more frequently and systematically screened for functional deficits, artificially inflating their SARC-F scores relative to controls. This limitation is now integrated into the revised Section 4.6.

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

I consider that the manuscript could be accepted.