IJPDS International Journal of Population Data Science

https://doi.org/10.23889/ijpds.v8i1.1768.review.r1.reva

Dear Editors of IJPDS, I would like to thank you for the opportunity to evaluate this manuscript.

I would like also to thank the authors for sharing their work for assessment. The manuscript is, in general, well written and easy to follow, relevant to the field, and well structured. It is alsowell organized.

The analysis and the methods used seem adequate to the aim of the study, as well as the results presented. Nevertheless, the paper has some points that need to be discussed and clarified. My general view is that the manuscript can be accepted after the corrections are addressed (accepted with major revisions (with further peer-review)). I hope my suggestions in the ‘Comments to the Author’ can contribute to its improvement.

TITLE AND ABSTRACT

The title is informative and captures well the content and breadth of the study (not misleading). The abstract is, in general, well structured and straightforward; essential details are presented. Nonetheless, there are a few suggestions for small changes in this section:

Background: there might be a typo in ‘studies’ (shouldn’t it be study’s aim?).

Results: it is not clear why the thousands’ separator is used for number of patients but not currency, please standardize it throughout the text.

Conclusions: I think that the phrase “Further research is needed to examine triggers for unplanned hospitalisation and to identify modifiable reasons for variation in hospital use among different cancer cohorts.” a little bit problematic. Although I agree more research is needed to understand in more depth the causes of hospitalisation in the last year of life, it might not be restricted only to unplanned stays, but all of them that we need to understand more. The same applies to ‘modifiable reasons’. It would be worth that more research is done to understand which factors influence in the resource use and length of stay for patients with different cancer types. And the words ‘modifiable reasons’ don’t sound adequate; so, please consider rewording the phrase.

INTRODUCTION (Background)

Overall, the introduction section is well written and straightforward. It builds a logical case and context for the problem statement. The research question is clear and concise, although the second part of the following statement could have its clarity improved: “and to question the extent to which current care pathways offer value”. It is not clear if this is indeed addressed in the results section of the study.

In addition, there are a few places where it is not clear:

- “Currently around 50% of people in Scotland die in hospital. [5, 6, 7]” – It is not clear if population in general or people diagnosed with cancer (focus of this study).

- “It is increasingly accepted that we should align clinical intervention with individual patients’ needs and preferences, moving away from the historic ‘doctor knows best’ culture. [11]” – the current wording gives the impression that the authors are making a value judgement; please consider rewriting this phrase.

- “The rising costs of cancer treatment, driven by new therapeutic options, is important context and necessitates that the true value of clinical interventions is understood so that scarce resources can be directed appropriately. [12, 13].” This phrase can be rewritten to increase clarity.

-“Further outcomes reported an exponential cost increase …”. Not clear what ‘further outcomes’ relate to.

- “… revealed a lack of health economic research. [15]” – Is the lack of health economic research specific to this topic?

- “…secondary healthcare use for patients with (advanced) cancer in their last year…” Not clear why ‘advanced’ is between brackets.

METHODS

The study adds to the literature already available on the subject. The research design is, in general terms well defined and clearly described. The data set and the population included in the analyses are described with sufficient detail and seem appropriate to the research question.

Nonetheless, there is one important issue that needs clarification in the text. From the description in the manuscript, the population included in the analyses “included everyone with a recorded cancer diagnosis in Scotland who died between 2012 and 2017 and were over 60 years of age on their date of death”. However, one of the inclusion criteria listed was “Healthcare data available for a minimum of 365 days prior to death” (which, according to Supplementary figure S1, “excluded if person died within 1 year from study start date of 1st Jan 2012”). In that sense, it seems that patients included in the analyses were in fact those who died from 01/01/2013 to 31/12/2017.

It is also not clear in the description what was the reason for choosing this specific timeframe (2012 to 2017) and the population over 60 years old at the time of death. In patient characteristics, it would be worth including the mean time from diagnosis as one of the patients characteristics’ variables presented, if available. Time from diagnosis, and therefore the “aggressiveness” of the cancer might be a prognosis factor for associated to costs at the last year of life for these patients.

In outcome measures, in the paragraph that details the health service costs from Scottish cost book, for readers outside Scotland, it is probably more important to detail which category included than the code, so one potential suggestion would be to change the wording to something similar to:

Scottish health service costs (Scottish cost book) were used to estimate the cost of inpatient care, mainly specialty group costs including inpatient data for: (i) all specialties excluding long stays (code R040), (ii) long stay specialties (R040LS), [21] and (iii) specialty group costs for day cases (R042). Costs related to critical care stays were included within mean costs.”

Please also clarify what “were included within mean costs” mean, if it was included in all categories (i to iii).

Please consider changing the wording for the ‘outpatient care’ subsection accordingly. In addition, please clarify what ‘nature of outpatient appointments per patient’ relates to.

Please also include what type of analyses were generated (it is not mentioned that results aggregated and by cancer type would be presented).

RESULTS

The results presented in the manuscript are comprehensive; they are presented effectively, making good use of tables and figures, including the material in the supplementary material. Results are organized in a way that is easy to understand; the results are contextualized by the authors.

One issue spotted in Tables 1, 3, 4 and all tables in the supplementary material is that the categories titles are missing from each of the categories presented in each of them (e.g. age category, sex, cause of death/type of cancer as cause of death, Comorbidity Index, Socioeconomic (SIMD) Index and Urban-rural indicator). In addition, in tables 3, 4 and all tables in the supplementary material, the table headers are missing (mean and range?, frequency, mean total costs?). Also, in Table 1 the order that the results are presented for the cause of death/type of cancer is not clear (it is not alphabetically or by decreasing frequency).

It might be also worth to provide (could be as a footnote) some explanation about the CI index.

In page 7 (about the results in table 2), the authors mention that “we observed differences in age at death across cancer types, with prostate cancer patients being oldest at their time of death”. It could be worth to include as part of the discussion what the potential impact could be of restricting the population age to 60+, if the authors believe there would be any (or in case the inclusion criteria is better detailed and justified in previous section this might not be needed).

In table 2 (or in the methods or table 1), it might be worth to clarify which categories for sex were used (as done for CI index in table 1). It is also unclear if presenting the results by cancer type as mean and sd for these categorical variables is the best way of presenting it (possibly the percentages in each category could be more meaningful for the reader).

As a last point, the text says that “Results from the univariate analysis (Supplementary Tables 2 to 6) reveal significantly lower costs associated with increased age, female gender and residing in the 3rd and 4th SIMD decile categories (some of the less deprived postcode areas).” However, results by sex are not presented in these tables.

The last phrase in the results section should be reworded (“Detailed results in Supplementary table 7 and in the supplementary figures 5 to 7.”).

DISCUSSION AND CONCLUSION

The discussion section is well written and great. Interpretations for the findings are appropriately presented, study limitations are considered in some depth, conclusions are well articulated, and recommendations for future studies is offered.

A few points that could be worth to mention:

- “This study confirms recent research showing that secondary care costs typically rise steeply in the last months of life. [4, 29]” Study 4, however, seems to be published by the same authors. It would be interesting (and necessary) to include more studies in this topic here.

- “An interesting finding in our study was the association between rurality and lower hospital costs, possibly reflecting proactive primary care for more rural populations, and alternative pathways to acute hospitalisation such as community hospital admission.” The lower costs in rural areas could be associated with lower access to hospital care in these areas, due to lower availability, more restricted transport or another factor? It would be interesting to include, if available, some literature here in the discussion about it (availability and access to cancer care in rural or more deprived areas).

- The exclusion of specialist palliative care from the analysis could be a big limitation of the analysis, since later stages of the disease could on one hand, increase the overall costs of care for patients before death, but on the other hand, could be proportionally less important than inpatient care for potentially being less technology-intensive. It is not clear as well if the costs included in the analysis include drug costs; the authors should clarify it in the methods section.

- “A further limitation arose from the way the CCI was derived. CCI values were linked solely to inpatient datasets (92% had at least one inpatient appointment) leaving 8% without a CCI value.” – The authors should include some explanation/more details on the methods section on CCI were derived/calculated.

-“Further data gaps related to specialist cancer treatments such as chemotherapy or radiotherapy.” – Apart from more details and clarification needed in the discussion related to this limitation of the study, the authors should make more clear in the methods section which costs are included in the categories listed (and which types of costs were not, like apparently chemotherapy or radiotherapy).

https://doi.org/10.23889/ijpds.v8i1.1768.review.r1.revb

The topic of the article is of public health importance, and the research has the potential to improve public health practice. However, I think the manuscript is not yet ready for publication. Please find my comments below.

(1) The abstract is incomplete, and there are inconsistencies between the abstract and the text. For example, there is a contradiction between the research question presented in the Abstract and the Background sections. In the Abstract section, the authors described that the study aimed to describe patterns of hospital-based healthcare use and associated costs in the last year of life for patients with a cancer diagnosis in Scotland. In the Background section, the aim is to understand more about the timing and nature of secondary healthcare use for patients with (advanced) cancer in their last year of life in Scotland, to identify factors associated with the variation, and to question the extent to which current care pathways offer value. Furthermore, there are results in the abstract not shown in the text and tables. Please, consider rewriting, including also the type of GLM model.

(2) The manuscript title is "Variation in hospital cost trajectories at the end-of-life by age, multimorbidity and cancer type." The study analyzed gender, age, the primary cause of death, comorbidity, rural-urban indicator, and the Scottish index of multiple deprivations (SIMD). Please, consider rewriting.

(3) The research design is insufficiently detailed to permit the study to replicate. For example, the methodology section does not explore the type of GLM models and the model' selection and validation process. Please consider detailing and describing data quality control.

(4) Results are incomplete and not organized in a way that is easy to understand. Statistical differences are not distinguished from meaningful differences. The manuscript explores cost values by type of care in the summary section (not shown in the text and tables) and presents a table with the costs stratified by cancer type and GLM variables. However, the authors did not explore these results. I suggest including Table 3 in the supplementary material to explore the costs by type of care. Please, consider exploring table 4 in the text.

(5) The sentence "(...) we observed differences in age at death across cancer types, with prostate cancer prostate patients being oldest at their time of death" is inconsistent with the information presented in table 2.

(6) Tables must be self-explanatory. Consider including titles and footnotes to explain all abbreviations. In table 1, consider including the variable name. The mean and standard deviation analysis of categorical variables sex and RU in table 2 was inappropriate. Consider specifying the x and y axes in figure 1. Are the data presented expressed in rates or absolute numbers? What is the currency analyzed in table 4?

(7) There is an inconsistency between the conclusions and the design, methods, and results. Please, consider articulating them.

(8) The literature is not critically appraised. I suggest comparing costs in the last year of life for cancer patients in Scotland with the other countries.

(9) Consider including an explicit statement of approval by an institutional review board (IRB) for studies directly involving human subjects.

(10) This paper is generally well written. Nevertheless, I recommend a revision for minor grammar and typographical errors (e.g., stomage/stomach; end of life/end-of-life; daycase/ day-case; day care/ day-case).

https://doi.org/10.23889/ijpds.v8i1.1768.r1.dec

Dear Katharina Diernberger, Xhyljeta Luta, Joanna Bowden, Joanne Droney, Elizabeth Lemmon, Giovanni Tramonti, Bethany Shinkins, Ewan Gray, Joachim Marti, Peter S. Hall:

We have reached a decision regarding your submission to International Journal of Population Data Science, "Variation in hospital cost trajectories at the end of life by age, multimorbidity and cancer type".

Please address the attached reviewers' comments and return to us: one clean and one tracked changes version of your revised manuscript, plus a point by point letter of response/rebuttal, by [editor to insert date here].

Our decision is to: Resubmit for Review

Kind Regards

We thank the editor and the reviewers for their time and thoughtful comments that have helped us to improve our manuscript. Our detailed responses below are included in bold directly underneath each Reviewer comment. Corresponding changes are made in the main body of the manuscript.

Reviewer A:

Dear Editors of IJPDS, I would like to thank you for the opportunity to evaluate this manuscript.

I would like also to thank the authors for sharing their work for assessment. The manuscript is, in general, well written and easy to follow, relevant to the field, and well structured. It is also well organized.

The analysis and the methods used seem adequate to the aim of the study, as well as the results presented. Nevertheless, the paper has some points that need to be discussed and clarified. My general view is that the manuscript can be accepted after the corrections are addressed (accepted with major revisions (with further peer-review)). I hope my suggestions in the ‘Comments to the Author’ can contribute to its improvement.

We thank the reviewer for their positive assessment as well as their detailed and valuable suggestions. 

TITLE AND ABSTRACT

The title is informative and captures well the content and breadth of the study (not misleading). The abstract is, in general, well structured and straightforward; essential details are presented. Nonetheless, there are a few suggestions for small changes in this section:

Background: there might be a typo in ‘studies’ (shouldn’t it be study’s aim?).

We have incorporated the reviewer’s suggestion.

Results: it is not clear why the thousands’ separator is used for number of patients but not currency, please standardize it throughout the text.

We have incorporated the reviewer’s suggestion.

Conclusions: I think that the phrase “Further research is needed to examine triggers for unplanned hospitalisation and to identify modifiable reasons for variation in hospital use among different cancer cohorts.” a little bit problematic. Although I agree more research is needed to understand in more depth the causes of hospitalisation in the last year of life, it might not be restricted only to unplanned stays, but all of them that we need to understand more. The same applies to ‘modifiable reasons’. It would be worth that more research is done to understand which factors influence in the resource use and length of stay for patients with different cancer types. And the words ‘modifiable reasons’ don’t sound adequate; so, please consider rewording the phrase.

We thank the reviewer for highlighting the exclusion of planned visits due to the wording. We agree that planned hospital based care activity is still a factor in an end of life care setting, though the last few month are typically characterized by an increase in unplanned hospital activity.

Changed the phrase “modifiable reasons” into “influenceable reasons for unwarranted…” as we are primarily interested in the reasons we can have an impact on, for example by changes to the service configuration.

INTRODUCTION (Background)

Overall, the introduction section is well written and straightforward. It builds a logical case and context for the problem statement. The research question is clear and concise, although the second part of the following statement could have its clarity improved: “and to question the extent to which current care pathways offer value”. It is not clear if this is indeed addressed in the results section of the study.

Thank you. Defining value in end of life care is a contentious issue therefore, we have removed this concept from the introduction to improve clarity.

In addition, there are a few places where it is not clear:

- “Currently around 50% of people in Scotland die in hospital. [5, 6, 7]” – It is not clear if population in general or people diagnosed with cancer (focus of this study).

We thank the reviewer for pointing this out. We have added a new sentence about in hospital deaths for the Scotland based cancer population.

- “It is increasingly accepted that we should align clinical intervention with individual patients’ needs and preferences, moving away from the historic ‘doctor knows best’ culture. [11]” – the current wording gives the impression that the authors are making a value judgement; please consider rewriting this phrase.

Thank you. It does indeed read that way – nevertheless it is the term used by Realistic Medicine. , hence why placed as a quote. We have re-written to make this clear: “It is  stated in the Scottish Chief Medical Officer report ‘Realistic Medicine’ that we should align clinical intervention with individual patients’ needs and preferences, moving away from the historic ‘doctor knows best’ culture. [11] “

- “The rising costs of cancer treatment, driven by new therapeutic options, is important context and necessitates that the true value of clinical interventions is understood so that scarce resources can be directed appropriately. [12, 13].” This phrase can be rewritten to increase clarity.

Thank you. Re-phrased sentence.

-“Further outcomes reported an exponential cost increase …”. Not clear what ‘further outcomes’ relate to.

Thank you. Clarified.

- “… revealed a lack of health economic research. [15]” – Is the lack of health economic research specific to this topic?

Thank you. Addressed

- “…secondary healthcare use for patients with (advanced) cancer in their last year…” Not clear why ‘advanced’ is between brackets.

Thank you. It was meant as in cancer and especially advanced cancer – which indeed is not necessary to point out as an end of life situation is coming with a high probability of the cancer being at an advanced stage so we have removed.

 METHODS

The study adds to the literature already available on the subject. The research design is, in general terms well defined and clearly described. The data set and the population included in the analyses are described with sufficient detail and seem appropriate to the research question.

Nonetheless, there is one important issue that needs clarification in the text. From the description in the manuscript, the population included in the analyses “included everyone with a recorded cancer diagnosis in Scotland who died between 2012 and 2017 and were over 60 years of age on their date of death”. However, one of the inclusion criteria listed was “Healthcare data available for a minimum of 365 days prior to death” (which, according to Supplementary figure S1, “excluded if person died within 1 year from study start date of 1st Jan 2012”). In that sense, it seems that patients included in the analyses were in fact those who died from 01/01/2013 to 31/12/2017.

This is an astute observation from the reviewer. We have corrected the study start date in the appendix to 1^st Jan 2011 which is the date from which we had healthcare data – one year before the first recorded death in the study.

It is also not clear in the description what was the reason for choosing this specific timeframe (2012 to 2017) and the population over 60 years old at the time of death. In patient characteristics, it would be worth including the mean time from diagnosis as one of the patients characteristics’ variables presented, if available. Time from diagnosis, and therefore the “aggressiveness” of the cancer might be a prognosis factor for associated to costs at the last year of life for these patients.

We agree that this would be a worthwhile additional characteristic, unfortunately we do not have access to the necessary data to make this addition.

In outcome measures, in the paragraph that details the health service costs from Scottish cost book, for readers outside Scotland, it is probably more important to detail which category included than the code, so one potential suggestion would be to change the wording to something similar to:

Scottish health service costs (Scottish cost book) were used to estimate the cost of inpatient care, mainly specialty group costs including inpatient data for: (i) all specialties excluding long stays (code R040), (ii) long stay specialties (R040LS), [21] and (iii) specialty group costs for day cases (R042). Costs related to critical care stays were included within mean costs.”

Thank you! We agree. Reviewers’ suggestion was used.

Please also clarify what “were included within mean costs” mean, if it was included in all categories (i to iii).

Thank you. Specified that!

Please consider changing the wording for the ‘outpatient care’ subsection accordingly. In addition, please clarify what ‘nature of outpatient appointments per patient’ relates to.

Thank you. Changed accordingly

Please also include what type of analyses were generated (it is not mentioned that results aggregated and by cancer type would be presented).

Thank you. Added

 RESULTS

The results presented in the manuscript are comprehensive; they are presented effectively, making good use of tables and figures, including the material in the supplementary material. Results are organized in a way that is easy to understand; the results are contextualized by the authors.

One issue spotted in Tables 1, 3, 4 and all tables in the supplementary material is that the categories titles are missing from each of the categories presented in each of them (e.g. age category, sex, cause of death/type of cancer as cause of death, Comorbidity Index, Socioeconomic (SIMD) Index and Urban-rural indicator). In addition, in tables 3, 4 and all tables in the supplementary material, the table headers are missing (mean and range?, frequency, mean total costs?). Also, in Table 1 the order that the results are presented for the cause of death/type of cancer is not clear (it is not alphabetically or by decreasing frequency).

Thank you for pointing this out. Tables in main paper and supplementary material were changed accordingly.

It might be also worth to provide (could be as a footnote) some explanation about the CI index.

We have added to the methods section: “Comorbidity was estimated using the Charlson Comorbidity Index (CCI), based on secondary care coding, which entailed a 5 year look back from patients’ first contact with secondary care using ICD-10 code lists developed by Public Health Scotland.”

In page 7 (about the results in table 2), the authors mention that “we observed differences in age at death across cancer types, with prostate cancer patients being oldest at their time of death”. It could be worth to include as part of the discussion what the potential impact could be of restricting the population age to 60+, if the authors believe there would be any (or in case the inclusion criteria is better detailed and justified in previous section this might not be needed).

Thank you. We added “ We were restricted to deaths in over 60 year olds due to the availability of data which was part of a wider research programme” to the limitations.

In table 2 (or in the methods or table 1), it might be worth to clarify which categories for sex were used (as done for CI index in table 1). It is also unclear if presenting the results by cancer type as mean and sd for these categorical variables is the best way of presenting it (possibly the percentages in each category could be more meaningful for the reader).

That is true! As not discussed in text anyway – deleted.

As a last point, the text says that “Results from the univariate analysis (Supplementary Tables 2 to 6) reveal significantly lower costs associated with increased age, female gender and residing in the 3rd and 4th SIMD decile categories (some of the less deprived postcode areas).” However, results by sex are not presented in these tables.

Added the table to the supplementary material and changed all subsequent numbering in the main paper.

The last phrase in the results section should be reworded (“Detailed results in Supplementary table 7 and in the supplementary figures 5 to 7.”).

Agree. Amended.

 DISCUSSION AND CONCLUSION

The discussion section is well written and great. Interpretations for the findings are appropriately presented, study limitations are considered in some depth, conclusions are well articulated, and recommendations for future studies is offered.

A few points that could be worth to mention:

- “This study confirms recent research showing that secondary care costs typically rise steeply in the last months of life. [4, 29]” Study 4, however, seems to be published by the same authors. It would be interesting (and necessary) to include more studies in this topic here.

Thank you for pointing that out. Two additional studies added.

- “An interesting finding in our study was the association between rurality and lower hospital costs, possibly reflecting proactive primary care for more rural populations, and alternative pathways to acute hospitalisation such as community hospital admission.” The lower costs in rural areas could be associated with lower access to hospital care in these areas, due to lower availability, more restricted transport or another factor? It would be interesting to include, if available, some literature here in the discussion about it (availability and access to cancer care in rural or more deprived areas).

Excellent idea, thank you. Added some more references.

- The exclusion of specialist palliative care from the analysis could be a big limitation of the analysis, since later stages of the disease could on one hand, increase the overall costs of care for patients before death, but on the other hand, could be proportionally less important than inpatient care for potentially being less technology-intensive. It is not clear as well if the costs included in the analysis include drug costs; the authors should clarify it in the methods section.

No drug costs and specialist palliative care were included due to non-availability of data. Added to the limitation section.

- “A further limitation arose from the way the CCI was derived. CCI values were linked solely to inpatient datasets (92% had at least one inpatient appointment) leaving  8% without a CCI value.” – The authors should include some explanation/more details on the methods section on CCI were derived/calculated.

We have added text to the methods section “Comorbidity was estimated using the Charlson Comorbidity Index (CCI), based on secondary care coding, which entailed a 5 year look back from patients’ first contact with secondary care using ICD-10 code lists developed by Public Health Scotland.”

-“Further data gaps related to specialist cancer treatments such as chemotherapy or radiotherapy.” – Apart from more details and clarification needed in the discussion related to this limitation of the study, the authors should make more clear in the methods section which costs are included in the categories listed (and which types of costs were not, like apparently chemotherapy or radiotherapy).

Added more information in the limitation section as well as some additional explanation on the Scottish cost book in the methods.

Recommendation: Revisions Required

------------------------------------------------------

------------------------------------------------------
Reviewer B:

Manuscript title: Variation in hospital cost trajectories at the end of life by age, multimorbidity and cancer type

Reviewer’s comments:

The topic of the article is of public health importance, and the research has the potential to improve public health practice. However, I think the manuscript is not yet ready for publication. Please find my comments below.

We thank the reviewer for their valuable suggestions. 

(1) The abstract is incomplete, and there are inconsistencies between the abstract and the text. For example, there is a contradiction between the research question presented in the Abstract and the Background sections. In the Abstract section, the authors described that the study aimed to describe patterns of hospital-based healthcare use and associated costs in the last year of life for patients with a cancer diagnosis in Scotland. In the Background section, the aim is to understand more about the timing and nature of secondary healthcare use for patients with (advanced) cancer in their last year of life in Scotland, to identify factors associated with the variation, and to question the extent to which current care pathways offer value. Furthermore, there are results in the abstract not shown in the text and tables. Please, consider rewriting, including also the type of GLM model.

Thank you. We amended the Abstract, which we believe is complete  we have made our best efforts to improve the consistency between the abstract and main text, included in various places as tracked changes.

(2) The manuscript title is "Variation in hospital cost trajectories at the end-of-life by age, multimorbidity and cancer type." The study analyzed gender, age, the primary cause of death, comorbidity, rural-urban indicator, and the Scottish index of multiple deprivations (SIMD). Please, consider rewriting.

It is true that other factors were included in the analysis, nevertheless the study’s main focus was on age, multimorbidity and cancer-type – hence we focus the title on these factors.

(3) The research design is insufficiently detailed to permit the study to replicate. For example, the methodology section does not explore the type of GLM models and the model' selection and validation process. Please consider detailing and describing data quality control.

Thank you for the observation, we have added more information about the type of GLM in the methods. Data quality control followed the well established internal protocols of Public Health Scotland, undertaken prior to our receiving the anonymised research extract – we have added this for clarity to the methods.

(4) Results are incomplete and not organized in a way that is easy to understand. Statistical differences are not distinguished from meaningful differences. The manuscript explores cost values by type of care in the summary section (not shown in the text and tables) and presents a table with the costs stratified by cancer type and GLM variables. However, the authors did not explore these results. I suggest including Table 3 in the supplementary material to explore the costs by type of care. Please, consider exploring table 4 in the text.

• Thank you. We have made our best efforts to articulate the finding of statistical and health service significance in the text. We believe that table three contains high priority results and belongs in the main paper. Additionally, Figure 1 represents the resource use by type of care; the corresponding figure presenting the costs can be found in the Appendix (Supplementary figure 4). The main share of the costs is spent as inpatient costs; therefore we believe that a detailed exploration by type of care would not add value. It was seen that some cancer types have more frequent use (therefore costs) of outpatient and day-case services, which is mainly to do with the required treatment intensity and service configuration of particular cancers - we have highlighted this in the discussion. The different cost and resource use patterns of the various cancer types are thought to be best presented in figures.
• Although we attempted to explore all results in the paper, we accept that the word limit has constrained us; There was indeed no reference to table 4 and we have now added that to the text.

 (5) The sentence "(...) we observed differences in age at death across cancer types, with prostate cancer prostate patients being oldest at their time of death" is inconsistent with the information presented in table 2.

Thank you for pointing this out. It is indeed breast cancer! Changed accordingly.

(6) Tables must be self-explanatory. Consider including titles and footnotes to explain all abbreviations. In table 1, consider including the variable name. The mean and standard deviation analysis of categorical variables sex and RU in table 2 was inappropriate. Consider specifying the x and y axes in figure 1. Are the data presented expressed in rates or absolute numbers? What is the currency analyzed in table 4?

Thank you.

• Titles for categories added, which explains the abbreviations.
• ad. table 2: Urban-rural indicator and sex were not explored in this context - deleted.
• Data presented in absolute numbers/per month – clarified in text.
• Data presented in absolute numbers/per month – clarified in text.
• Currency throughout the paper is £ -added to the caption of table.

(7) There is an inconsistency between the conclusions and the design, methods, and results. Please, consider articulating them.

We have made our best efforts to improve the consistency between the conclusions, design, methods and results – included as tracked changes.

(8) The literature is not critically appraised. I suggest comparing costs in the last year of life for cancer patients in Scotland with the other countries.

The focus of the literature search prior to our analysis was on UK based studies as the findings are very much linked to UK service configuration, which are health system specific.

(9) Consider including an explicit statement of approval by an institutional review board (IRB) for studies directly involving human subjects.

Thank you for pointing this out. Ethics statement included.

(10) This paper is generally well written. Nevertheless, I recommend a revision for minor grammar and typographical errors (e.g., stomage/stomach; end of life/end-of-life; daycase/ day-case; day care/ day-case).

We have taken reviewer’s comment in full consideration. A professional English native speaker corrected our manuscript. Corrections are highlighted in track changes.

https://doi.org/10.23889/ijpds.v8i1.1768.review.r2.reva

Dear Editors of IJPDS,

I would like to thank you for the opportunity to evaluate this manuscript again, and the authors for working on the suggestions from both reviewers. The new version of the manuscript reflects the improvements made, nonetheless, there are still a few minor points that need to be addressed:

- Slight mismatch between the study’s aim as described in the abstract and main text has been addressed, but still persists. Abstract mentions ‘hospital-based healthcare use and associated costs’, whilst main text only mentions ‘resource use’. The main text also mentions ‘and to identify factors associated with the variation’, but it is not clear where variations would be expected to be seen (e.g., between cancer types, between years, socio-economic groups?).

I also agree with the other reviewer that some of the results mentioned in the abstract are only presented in supplementary table, but not in the text. This issue has not been addressed by the authors, and I would recommend revising the consistency between abstract, results text and discussion a bit further.

- Tiny typo in ‘Direct costs include e all bed days’ in page 5.

- Regarding the results from Table 2, although I appreciate the authors removing the results from the table for categorical variables which were more difficult to interpret, the accompanying text hasn’t been modified, and RU results are still mentioned in “There were noticeable differences in the socioeconomic (SIMD) and rural-urban status of patients by cancer type.” and in “There was also variation in rural-urban status of patients by cancer type.” I am not entirely sure if just removing it from this table would be the best solution (since it is still mentioned in the methods and in other parts of the text, and are included in the GLM models). Please consider what the best approach would be to deal with this issue.

- Supplementary tables: category titles still missing in Table 10

- The manuscript has improved, nonetheless I recommend a revision before final submission for minor grammar and typographical errors, especially for additional spaces or spaces bring removed with the last edit.

https://doi.org/10.23889/ijpds.v8i1.1768.review.r2.revb

The manuscript has been greatly improved; however, some issues remain.

1)Abstract:

The results "Mean total inpatient, outpatient and day-case costs per patient were £10261, £1275 and £977 respectively" were not presented in the Results section, so I suggest omitting them from the abstract.

2)Methods (Supplementary):

- Figure S1: Please include the number of records for all box presenting records excluded from the analysis (e.g., enrolment days <365)

3)Results:

- Please review the text “were aged between 70 and 84 years at time of death”. According to the table 1, the correct is “were aged between 70 and 79 years at time of death”.

- Table1 - Please review the percent figures, as some variables did not add up to 100%. For instance, age category (99,99%), Scottish Index of Multiple Deprivation (99,99%) and Urban-rural indicator (99,71%). Please indicate the occurrence of missing data (e.g., Scottish Index of Multiple Deprivation) in the footnote.

- Please omit comments about the variable rural-urban status as it was removed from table 2. For instance:

"There were noticeable differences in the socioeconomic (SIMD) and rural-urban status…".

"There was also variation in rural-urban status of patients by cancer type."

- I could not identify in Table 4 the data supporting the affirmative "In our study population, older age was negatively correlated with a higher comorbidity burden (Table 4)."

- Please review all figures and tables titles to make them more informative, include in the footnote’s explanations about abbreviations, and present all units of measurement and labels of the x and y axes.

- Please refer to the appropriate part of the Results sections text the Supplementary Figure 3.

- Please, in Table 4, replace stomage for stomach.

- Please cite the interaction term used in Supplementary Figures 5.

https://doi.org/10.23889/ijpds.v8i1.1768.r2.dec

Dear Katharina Diernberger, Xhyljeta Luta, Joanna Bowden, Joanne Droney, Elizabeth Lemmon, Giovanni Tramonti, Bethany Shinkins, Ewan Gray, Joachim Marti, Peter S. Hall:

We have reached a decision regarding your submission to International Journal of Population Data Science, "Variation in hospital cost trajectories at the end of life by age, multimorbidity and cancer type".

Please address the attached reviewers' comments and return to us: one clean and one tracked changes version of your revised manuscript, plus a point by point letter of response/rebuttal, by [editor to insert date here].

Our decision is to: Resubmit for Review

Kind Regards