Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript presented by Kua et al. addresses an important topic, particularly within the context in which the authors operate. Building on this aspect, in order to position their work within a broader framework of interest for the international scientific community, not limited to the Asian context, I would suggest that the authors reconsider the title and restructure the presentation of their results around a more general and widely relevant issue, namely the importance of assessing REE for nutritional management and the inadequacy of currently available predictive equations. The specific Asian context, including its particular characteristics and the focus on hospitalized elderly subjects, could then be presented in the final part of the manuscript. Otherwise, the topic may appear overly geographically restricted in scope and relevance.

In the current manuscript, the authors report a considerable amount of data that has already been presented in their previous publication, and this should be revised. It would be sufficient to briefly summarize only the key methodological aspects, while referring readers to the previous manuscript for a detailed description, as the authors themselves indicate. This applies particularly to the Results section: there is no need to repeat information such as the number of subjects invited to participate, those enrolled, etc.

One aspect that generates some confusion is the fact that the authors state they have already modified and adapted the predictive equations, yet they present this modification as the novelty of the current manuscript. This point should be clarified.

Overall, the Results section is not entirely easy to read, as it is highly technical. I would recommend synthesizing the findings and integrating them with a clearer interpretation of the statistical analyses performed, presenting the results in a way that enhances their practical relevance and applicability.

Finally, the English language requires linguistic revision, as in some sections the intended meaning is unclear. For example, in line 94 (“was considered…”), the sentence is difficult to understand as currently formulated and may even be misleading.

A statistician should carefully review the statistical analyses and the interpretation of the results, as the authors have performed complex and potentially impactful analyses that require thorough methodological evaluation which I can not provide.

Comments on the Quality of English Language

Linguistic revision is needed.

Author Response

We thank the reviewer for his/her insightful and constructive feedback on our manuscript. We have carefully considered each point and have revised the manuscript accordingly (in blue) to improve its clarity, focus and clinical relevance.

1. On Generalization and the Asian Context:

• Reviewer Suggestion: Restructure to move the specific Asian context to the end to broaden international interest.
• Author Response: We appreciate the suggestion to contextualize our findings within a broader global framework. The Introduction has been revised to emphasize the universal inadequacy of standard predictive equations (PEs) in geriatric populations with multimorbidity. However, we have maintained the multi-ethnic Asian cohort as the primary focus because it represents a distinct metabolic phenotype. Highlighting these population-specific discrepancies specifically the systematic error driven by demographic and body composition variance is essential for demonstrating why "universal" equations frequently fail in specific clinical contexts.

2. On Redundancy and Repeating Previous Publication Data:

• Reviewer Suggestion: Summarize methodology and remove repeated result data (enrolment numbers, etc.).
• Author Response: We agree with the reviewer’s assessment. The Materials and Methods and Results sections have been significantly streamlined. Detailed recruitment and enrolment data have been replaced with a concise summary and a direct reference to the derivation study. This allows the current manuscript to focus exclusively on the recalibration strategy and validation metrics. 

3. On Clarifying the Novelty (Recalibration vs. Development):

• Reviewer Suggestion: Clarify the difference between previous adaptations and the current manuscript.
• Author Response: We apologize for the confusion. We have refined the text to explicitly distinguish between the de novo development of the SOAR equations in our prior work and the current objective: the recalibration of commonly used PEs. The novelty of this study lies in the application of BMI-stratified, slope-only adjustment factors to enhance the accuracy of these widely used tools for the Asian clinical setting.

 

4. On Readability and Statistical Interpretation of Results:

• Reviewer Suggestion: Synthesize technical findings and enhance practical relevance.
• Author Response: The Results and Discussion have been restructured to prioritize bedside application. We have transitioned to a methodology utilizing actual body weight across all BMI categories, aligned with ESPEN 2022 practical guidelines. This is supported by a new recommendation framework (Table 6) and BMI-stratified multipliers that provide a simplified, evidence-based solution for mitigating the risks of overfeeding and malnutrition.

5. On Language and the Rephrasing of Line 94:

• Reviewer Suggestion: Revise confusing phrasing, specifically "was considered..."
• Author Response: We appreciate this correction. The sentence in question (formerly line 94, now line 101-103) has been rephrased for clarity and precision: “Predictions were classified as underpredictions or overpredictions if the predicted REE deviated from the measured value by more than 10% in either direction".

6. On Statistical Review:

• Reviewer Suggestion: A statistician should review the complex analyses.
• Author Response: We thank the reviewer for this important suggestion. We agree that the complexity of recalibrating predictive equations requires rigorous statistical oversight. Consequently, our statistician has re-run and updated all analyses presented in this revised manuscript. The analytical framework was strengthened to include sensitivity analyses for proportional bias, BMI-stratified slope-only recalibration, and extensive internal validation through bootstrap resampling. We believe these updates provide a more robust and clinically applicable model for estimating energy expenditure in this specific population.

Additional Changes:

1. Statistical and Methodological Enhancements

To strengthen the manuscript’s statistical rigor, we incorporated a formal sensitivity analysis to assess proportional bias. This allows for a precise quantification of how predictive errors scale linearly as BMI diverges from the median, ensuring a transparent evaluation of model performance at physiological extremes. Additionally, our analytical approach was refined by utilizing multiple linear regression on optimal steady-state data to establish a robust baseline prior to the derivation of BMI-specific recalibration factors.

2. Population Re-stratification and Clinical Context

We re-stratified the BMI categories to more accurately represent the distribution of our 400 participants, which now includes an expanded "Normal Weight" cohort of 203 individuals. To provide deeper clinical context as suggested by the other reviewer, we introduced a detailed comorbidity prevalence table (Table 2). This table characterizes the cohort across 11 common conditions (including cancer, orthopaedic surgery, and cardiovascular diseases) stratified by weight status, highlighting the metabolic complexity of this geriatric population.

3. Model Refinement and Performance Benchmarking

The primary predictive tool has been explicitly defined as the New SOAR PE 1:

963.67 + (8.56 x weight) - (5.6 x age)

In response to reviewer feedback, we expanded the Results section to include a direct performance comparison between traditional equations utilizing Adjusted Body Weight (ABW) and our recalibrated formulas using actual body weight. This benchmarking confirms that our population-specific slope adjustments yield significantly higher predictive precision than historical weight-adjustment methods.

4. Practical Application and Guidelines Alignment

Finally, we have enhanced the practical utility of our findings by translating technical results into a clear decision-support framework. A new recommendation table (Table 6) has been added, pairing each BMI category with a specific suggested formula and its clinical rationale. This shift aligns our methodology with the latest ESPEN practical guidelines, providing clinicians with a simplified, evidence-based "plug-and-play" workflow for bedside nutritional management.

Reviewer 2 Report

Comments and Suggestions for Authors

This study evaluated the performance of seven established equations among 400 hospitalized Asian older adults using indirect calorimetry (IC). To enhance predictive reliability, an adjustment was applied to account for variations across body mass index groups. This is a continuation of a previous study of the same group, in which they established several novel prediction equations (PEs) with superior accuracy for multi-ethnic Asian older adults with complex comorbidities than the occidental formulas.

 

It is suitable for publication, but it needs several improvements.

 

• The authors have to add the comorbidities of the patients in order to better characterize the population included in the study. They should specify if they have an oncological process or a surgical procedure and the length of stay at hospital.
• In real clinical practice, in order to take in account, the limitations of the formulas in subjects with extreme BMIs, we use adjusted weight. The authors might talk about this concept in the “Discussion”, as well as do a comparison between the REE obtained with this adjusted weight and the obtained with de novel formula with the new coefficient, in the “Results”.
• The authors might add an infographic with the formulas that they propose to use in daily clinical practice, for better spread their results.

Author Response

We thank the reviewer for his/her constructive feedback and for recognizing the suitability of our study for publication. We have addressed each of the points raised to better characterize our study population and enhance the clinical utility of our findings. The changes to the manuscript are in blue.

1. Characterization of the Population (Comorbidities and LOS)

• Reviewer Comment: “The authors have to add the comorbidities of the patients in order to better characterize the population included in the study. They should specify if they have an oncological process or a surgical procedure and the length of stay at hospital.”

• Author Response: We agree that detailing the clinical profile of our participants is essential for contextualizing the results. We have added a new Table 2: Table 2: Prevalence of Common Comorbidities Across BMI Categories. which stratifies conditions including diabetes, cardiovascular disease, cancer/suspected cancer, and bone surgery/fractures. Regarding LOS, as this was a cross-sectional study designed to calibrate equations during steady-state recovery, LOS was not systematically captured. We have acknowledged this as a limitation in the revised Discussion to ensure transparency.

 

2. Integration and Comparison of Adjusted Body Weight (ABW)

• Reviewer Comment: "In real clinical practice, in order to take in account, the limitations of the formulas in subjects with extreme BMIs, we use adjusted weight. The authors might talk about this concept in the “Discussion”, as well as do a comparison between the REE obtained with this adjusted weight and the obtained with de novel formula with the new coefficient, in the “Results”."
• Author Response: This is a pertinent suggestion that reflects traditional clinical practice. We have updated Table 5(d) to include a direct performance comparison between conventional equations using Adjusted Body Weight (ABW) and our recalibrated models (R*) utilizing actual body weight. Our findings demonstrate that the recalibrated actual body weight model (Accuracy: 57%) matches or exceeds the performance of standard ABW applications (Accuracy: 52%), while being significantly more efficient for bedside use. We have further justified this shift by citing the ESPEN 2022 guidelines, which find no clinical necessity for ABW in older adults.

3. Clinical Infographic

• Reviewer Comment: "The authors might add an infographic with the formulas that they propose to use in daily clinical practice, for better spread their results."
• Author Response: To facilitate rapid clinical adoption, we have synthesized our recommendations into Table 6: Recommended Predictive Equations for REE Across BMI Categories. This table functions as a practical decision tree, guiding clinicians to the optimal predictive equation specifically the New SOAR PE 1 (963.67 + 8.56 x weight - 5.6 x age) for the underweight, normal-weight, and obese cohorts, or SOAR PE 2 for patients in the overweight range. By consolidating these findings into a concise, actionable framework, we aim to reduce the cognitive burden on healthcare professionals and ensure consistent, evidence-based nutritional care at the bedside.

Additional Changes:

In this updated version of the manuscript, we have comprehensively revised the Abstract, Methodology, Results, Discussion, and Conclusion to reflect significant analytical improvements.

We have strengthened the statistical rigor of the study by incorporating a formal sensitivity analysis to assess proportional bias. This addition quantifies how predictive errors scale linearly as BMI diverges from the median, offering a more transparent evaluation of model performance at physiological extremes. Furthermore, we refined our analytical framework by applying multiple linear regression to optimal steady-state data, establishing a robust baseline prior to the derivation of BMI-specific recalibration factors.

Regarding the study population and predictive modelling:

• Population Re-stratification: We updated the BMI categories to better represent the distribution of our 400 participants, which now includes an expanded "Normal Weight" cohort of 203 individuals.
• Model Specification: We have explicitly defined the New SOAR PE 1 equation as: 963.67 + (8.56 x weight) - (5.6 x age)

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have improved the corrections mentioned. The manuscript is suitable for publication in this way.