Comparison of Nutrition Risk Screening 2002 and Subjective Global Assessment Form as Short Nutrition Assessment Tools in Older Hospitalized Adults

Round 1

Reviewer 1 Report

1. Please omit results from the methods section. For example, the authors describe reasons for hospitalization. This information is appropriately included in the Results section.
2. The authors omitted approximately 1/3 of all patients due to missing data. By this do the authors mean that any missing data resulted in exclusion? Please clarify.
3. The information regarding patient inclusion can be presented as a flow diagram in which the number of patients excluded for each reason is presented. Please see the attached file.
4. In the statistical analysis, the authors state that variable distribution was tested for normality. Please state the method used
5. The authors used non-parametric tests to estimate correlations and to detect differences. Did all of the variables have distributions significantly deviating from normal?
6. Please substitute the term, “weight” for “body mass”
7. Tables 2-4 present data by sex. Since a by-sex difference is not a stated study question, it is less cumbersome to present the data for the entire study population.
8. Tables 5 and 6 present data by nutrition risk status. This is very didactic but could be combined into a single table to facilitate the comparison between the two methods of assessing nutrition risk.
9. The authors do not appear to directly compare the two methods in any of the analyses, this should be added since this is the stated purpose of the study. Despite this, the authors state that the NRS better discriminates for elevated nutrition risk. How did the authors arrive at this statement? Please explain.
10. The sensitivity and specificity for NRS and SGA are presented in the Discussion Section but this was not mentioned in the Methods Section. Please present this data in the Methods Section.

Comments for author File: Comments.docx

Author Response

We would like to thank the Reviewers for all the constructive comments.

Reviewer 1.

Comments and Suggestions for Authors

1. Please omit results from the methods section. For example, the authors describe reasons for hospitalization. This information is appropriately included in the Results section.

The characteristics of the patients have been moved from the Methods to the Results section as suggested.

1. The authors omitted approximately 1/3 of all patients due to missing data. By this do the authors mean that any missing data resulted in exclusion? Please clarify.

The study was conducted in the “real world” geriatric ward. Therefore, many subjects were not able to answer the questions (severe dementia, terminal illness, palliative state). Because of the incomplete data those patients were excluded from the study. This has been clarified in the Methods section.

1. The information regarding patient inclusion can be presented as a flow diagram in which the number of patients excluded for each reason is presented. Please see the attached file.

Flow diagram has been added as suggested. The number of patients excluded for each reason has been presented.

1. In the statistical analysis, the authors state that variable distribution was tested for normality. Please state the method used

Kolmogorov-Smirnov test was used for normality check and Levene's test for equality of variances check. This information has been added.

1. The authors used non-parametric tests to estimate correlations and to detect differences. Did all of the variables have distributions significantly deviating from normal?

In fact, not all the variables had distributions deviating from normal. We used Spearman’s correlation coefficients for the unity of data presentations. Furthermore, the results of Pearson’s correlations were very similar. This information has been added to the results section.

1. Please substitute the term, “weight” for “body mass”

Has been corrected.

1. Tables 2-4 present data by sex. Since a by-sex difference is not a stated study question, it is less cumbersome to present the data for the entire study population.

We believe that the presentations of the results both in the whole studied population and separately in women and men provide some additional information. The fact that the distribution of both NRS-2002 and SGA was very similar in women and men, and correlations of NRS-2002 and SGA with age, anthropometric data and other CGA tools were generally similar in both sexes provides important practical information about usefulness of both nutritional tools equally in older women and men. This data about lack of by-sex difference has been provided and discussed more unequivocally.

1. Tables 5 and 6 present data by nutrition risk status. This is very didactic but could be combined into a single table to facilitate the comparison between the two methods of assessing nutrition risk.

Thank you for this very practical comment. Indeed, after merging both tables the data seems to be presented more clearly.

1. The authors do not appear to directly compare the two methods in any of the analyses, this should be added since this is the stated purpose of the study. Despite this, the authors state that the NRS better discriminates for elevated nutrition risk. How did the authors arrive at this statement? Please explain.

In our study, the prevalence of malnutrition was 22% according to NRS-2002 and 15% according to SGA. Several studies comparing malnutrition short assessment tools used SGA as a reference method or a “gold standard”. In the present study, the agreement between the two short nutrition assessment tools was very high. The sensitivity of NRS-2002 to detect malnutrition was 77.4% and specificity was 87.7% as compared to SGA. Relatively few studies assessed malnutrition measures in relation to the CGA measurements in older subjects. Both NRS-2002 and SGA showed correlation with anthropometric data and CGA measurements concerning physical and cognitive functioning. Significant association of both tools was also observed with VES-13. Lack of correlation with age and weaker correlations with physical, cognitive and frailty data for SGA may suggest that NRS-2002 might be more suitable for hospitalized older adults. This information has been provided more clearly in the Results and the Discussion.

1. The sensitivity and specificity for NRS and SGA are presented in the Discussion Section but this was not mentioned in the Methods Section. Please present this data in the Methods Section.

As stated above, the comparison between the two methods has been provided. The sensitivity and specificity for NRS and SGA have been presented in the Methods section as suggested:

“The sensitivity (the proportion of SGA B+C cases correctly identified as NRS-2002 3-7 cases) and specificity (the proportion of SGA A correctly identified as NRS-2002 0-2 cases) of NRS-2002 to detect malnutrition as compared to SGA was calculated.”

Reviewer 2 Report

The study done by Kroc et al. was to compare Nutrition Risk Screening 2002 (NRS-2002) and Subjective Global Assessment Form (SGA) with Comprehensive Geriatric Assessment (CGA) measurements in older hospitalized adults. There are few comments as following:

1. Introduction: Authors used the Comprehensive Geriatric Assessment (CGA) as the standard tool in this study. Authors should briefly introduce the CGA, and explain why this tool was used as the standard tool in this study.
2. Design of the study and participants: Several tests (i.e., NRS-2002, SGA, ADL, IADL, MMSE, GDS, VES-13) were performed in this study. However, it is not clear who conducted these tools, and when (i.e., at the admission day or any day during hospitalization) these tools were conducted.
3. Statistical analysis: Student’s t-test or Mann-Whitney Rank Sum Test rather than ANOVA is recommended to compare the variable differences between two groups.
4. Table 1: There is a little bit confused about body mass (kg). Is body mass equal to body weight?
5. Discussion, line 208-210: Authors stated that the sensitivity of NRS-2002 to detect malnutrition was 77.4% and specificity was 87.7% as compared to SGA in the present study. However, the reviewer had no idea how the sensitivity and specificity were calculated since these results were not mentioned in the result section.

Author Response

We would like to thank the Reviewers for all the constructive comments.

Reviewer 2.

Comments and Suggestions for Authors

The study done by Kroc et al. was to compare Nutrition Risk Screening 2002 (NRS-2002) and Subjective Global Assessment Form (SGA) with Comprehensive Geriatric Assessment (CGA) measurements in older hospitalized adults. There are few comments as following:

1. Introduction: Authors used the Comprehensive Geriatric Assessment (CGA) as the standard tool in this study. Authors should briefly introduce the CGA, and explain why this tool was used as the standard tool in this study.

CGA was briefly introduced and the rationale about its use was presented in the Introduction and the Discussion.

“In older adults with multiple deficiencies and comorbidities, the routine format of medical examination and other common tests and procedures is usually not sufficient. Therefore, the Comprehensive Geriatric Assessment (CGA) has been developed to address patients’ problems with medical comorbidities, functional status and psychosocial capacities. CGA is a multidisciplinary set of procedures that identifies medical, psychosocial, and functional capabilities of an older adult. CGA is a standard assessment methodology at geriatric wards.”

1. Design of the study and participants: Several tests (i.e., NRS-2002, SGA, ADL, IADL, MMSE, GDS, VES-13) were performed in this study. However, it is not clear who conducted these tools, and when (i.e., at the admission day or any day during hospitalization) these tools were conducted.

All the tests were conducted by the physicians of the geriatric ward at admission. This additional information has been provided.

1. Statistical analysis: Student’s t-test or Mann-Whitney Rank Sum Test rather than ANOVA is recommended to compare the variable differences between two groups.

The one-way analysis of variance (ANOVA) is an equivalent of Student’s t-test for comparison of numeric data in two groups. The p-values are the same.

1. Table 1: There is a little bit confused about body mass (kg). Is body mass equal to body weight?

Has been corrected to ”body mass”.

1. Discussion, line 208-210: Authors stated that the sensitivity of NRS-2002 to detect malnutrition was 77.4% and specificity was 87.7% as compared to SGA in the present study. However, the reviewer had no idea how the sensitivity and specificity were calculated since these results were not mentioned in the result section.

Several studies comparing malnutrition short assessment tools used SGA as a reference method or a “gold standard”. In the present study, the agreement between the two short nutrition assessment tools was very high. The sensitivity of NRS-2002 to detect malnutrition was 77.4% and specificity was 87.7% as compared to SGA.

The calculation methodology has been presented in the Methods section and comparisons given in the Results section of the study.

Reviewer 3 Report

This paper reports the results of the comparison of two widely recommended short nutrition assessment tools -NRS-2002 and SGA with CGA. In conclusion, the authors recommend using both NRS2002 and SGA to detect malnutrition or risk of malnutrition. Article introductions, materials and methods are well described. However, the discussion part needs to be further improved.

1. I agree with the author's opinion. Because what we have to avoid is that patients are overlooked in screening despite being undernourished.

However, I think the conclusion that the author recommends both is lacking in explanation.

The authors did not explain the gap between the prevalence of malnutrition 22% according to rate NRS-2002 and 20% according to SGA. Please show the sensitivity and specificity of the two indicators and explain why you recommend using both NRS2002 and SGA.

2. In 2018, GLIM criteria, a global diagnostic standard for undernutrition, was released. The authors did not mention this. Please also add it to the discussion section on the contribution of this result to GLIM criteria.

Author Response

We would like to thank the Reviewers for all the constructive comments.

Reviewer 3.

Comments and Suggestions for Authors

This paper reports the results of the comparison of two widely recommended short nutrition assessment tools -NRS-2002 and SGA with CGA. In conclusion, the authors recommend using both NRS2002 and SGA to detect malnutrition or risk of malnutrition. Article introductions, materials and methods are well described. However, the discussion part needs to be further improved.

1. I agree with the author's opinion. Because what we have to avoid is that patients are overlooked in screening despite being undernourished.

However, I think the conclusion that the author recommends both is lacking in explanation.

The authors did not explain the gap between the prevalence of malnutrition 22% according to rate NRS-2002 and 20% according to SGA. Please show the sensitivity and specificity of the two indicators and explain why you recommend using both NRS2002 and SGA.

Thank you for these comments. According to the suggestions of the Reviewer we have tried to justify the conclusions more in-depth. In fact, even higher difference in the prevalence of malnutrition between different nutritional tools has been reported in previous studies. In our study, the prevalence of malnutrition was 22% according to NRS-2002 and 15% according to SGA. Several studies comparing malnutrition short assessment tools used SGA as a reference method or a “gold standard”. In the present study, the agreement between the two short nutrition assessment tools was very high. The sensitivity of NRS-2002 to detect malnutrition was 77.4% and specificity was 87.7% as compared to SGA. Likewise, both nutritional scales at given cut-off points similarly discriminated anthropometric data and other CGA tools in the populations of well-nourished vs malnourished hospitalized older subjects. On the other hand, lack of correlation with age and weaker correlations with physical, cognitive and frailty data for SGA may suggest that NRS-2002 might be more suitable for hospitalized older adults. This potential disparity should be corroborated in future prospective studies.

1. In 2018, GLIM criteria, a global diagnostic standard for undernutrition, was released. The authors did not mention this. Please also add it to the discussion section on the contribution of this result to GLIM criteria.

This citation has been added to the discussions and potential contribution of present results to GLIM criteria has also been shortly discussed:

“Adding those physical, cognitive and frailty data to phenotypic and etiologic criteria of malnutrition proposed by the Global Leadership Initiative on Malnutrition (GLIM) might also have enriched diagnosis and severity grading of malnutrition.”

Round 2

Reviewer 2 Report

No additional comments