Sex-Based Analysis of Quality Indicators of End-of-Life Care in Gastrointestinal Malignancies

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Lees C, Seow H, Chan KKW et al. Sex-based analysis of quality indicators

Of end-of-life care in gastrointestinal malignancies.

            Acceptance of this manuscript is recommended with the three changes suggested in italics.

            The authors did a very careful retrospective study of the aggressive vs. comfort

EOL care offered to 29,529 women and 40,454 men in Ontario suffering from GI

cancers between 2006 and 2018.  Covariates used in the study were age at death, Deyo-Carlson Comorbidity Index,  cancer type. income quintile, and rurality.  The authors conclude that “significantly more males experienced aggressive EOL care when compared with females.”  Using their data, however, I believe this conclusion is over-stated, given their finding that of “ an association between younger age and increase odds of aggressive care in females, yet not males.”  In addition, the fact that the number of elderly women exceeded the number of elderly men is important given their finding that aggressive care diminished as a function of patient age.  Moreover, though not discoverable in this data set, the fact that there are fewer likely-to-

succeed treatment options for cervical and ovarian cancer vs. prostate cancer would seem to suggest that women were more likely than men to choose comfort care only.  Accordingly, it would be more accurate to say that no compelling differences were found in the scope of care offered to men and women which, in itself, is an important finding.

       Perhaps the most important finding in this research is the finding that ‘While rates of supportive care at EOL are increasing over time for both sexes, rates of aggressive care remain stable” in the Canada’s most populous province.  This finding merits and its implications merit stronger emphasis

Given the finding in numerous other studies that approximately three fourths of patients who complete advance directives choose limited or comfort care, by implication the authors’ finding that aggressive caré was more common at lower income levels and in rural areas suggests that advance care planning resources may be insufficient for patients in these groups.  Related to this, it is worth noting that this retrospective study did not evaluate the congruence between patients’ goals and the EOL treatment they received.  The discussion could be expanded to include these points.

Author Response

Thank you very much for the time spent on your thoughtful review. Please see our responses below:

1. The authors conclude that “significantly more males experienced aggressive EOL care when compared with females.”  Using their data, however, I believe this conclusion is over-stated, given their finding that of “an association between younger age and increase odds of aggressive care in females, yet not males.”  In addition, the fact that the number of elderly women exceeded the number of elderly men is important given their finding that aggressive care diminished as a function of patient age.  Moreover, though not discoverable in this data set, the fact that there are fewer likely-to-succeed treatment options for cervical and ovarian cancer vs. prostate cancer would seem to suggest that women were more likely than men to choose comfort care only.  Accordingly, it would be more accurate to say that no compelling differences were found in the scope of care offered to men and women which, in itself, is an important finding.

Thank you for raising these points. We do stand behind our conclusion that “significantly more males experienced aggressive EOL care when compared with females,” as our descriptive statistics demonstrate (19.0% vs 14.3%, SD = 0.13), while we have no data to support what care was offered versus what care was actually received. The multivariate logistic regression analysis simply shows that in the female population, younger age was associated with increased odds of aggressive care, while in the male population, no such association was found. However, we do appreciate the point that there are material differences between male and female populations with respect to age and subtypes of GI malignancies, which is commented in the discussion section. Accordingly, these were included as predictors within our multivariable logistic regression analysis.

2. Perhaps the most important finding in this research is the finding that ‘While rates of supportive care at EOL are increasing over time for both sexes, rates of aggressive care remain stable” in the Canada’s most populous province.  This finding merits and its implications merit stronger emphasis.

This is indeed an important point, though one we emphasized during our previous publication. Within this secondary analysis, our focus was instead on sex-based differences in quality of end-of-life care. Accordingly, we mainly commented on the lack of significant differences between sexes over time, rather than the trends found.

3. Given the finding in numerous other studies that approximately three fourths of patients who complete advance directives choose limited or comfort care, by implication the authors’ finding that aggressive care was more common at lower income levels and in rural areas suggests that advance care planning resources may be insufficient for patients in these groups.  Related to this, it is worth noting that this retrospective study did not evaluate the congruence between patients’ goals and the EOL treatment they received.  The discussion could be expanded to include these points.

Thank you for this point. We have added to our discussion accordingly, noting that, “nor did this study investigate congruence between advance care planning and the end-of-life care received.”

Reviewer 2 Report

Comments and Suggestions for Authors

This study uses information on patients with gastrointestinal (GI) malignancies and the end of life care (EOL) they received during their last 14-30 days of life in Ontario, between 2006 and 2018, linking registry and administrative data, to explore “sex-based quality of EOL care”. The following points will help strengthen the manuscript.

1. Lines 30-54: Could authors add here to explain which studies fall under what they describe as “disparities according to sex have been previously described…” in the Abstract? What were the gaps they identified in those studies? How does the current study fill those gaps?
2. Lines 56-115: Could authors add a Settings section explaining the healthcare framework within which their initial and present “population-based, retrospective cohort stud(dies)…” were performed?
3. Lines 56-115: Given that the study time horizon spans more than a decade, please explain the changes in health care services/public policy and approaches regarding EOL care and the approaches to the treatment for GI malignancies that might have an impact on the ‘quality of EOL care’.
4. Lines 56-115: Please consider providing the information contained in this section broken down into sub-sections such as setting, data sources, study population… ethical considerations…
5. Lines 56-57: Could authors explain what they mean by “secondary analysis”? Are authors using the same database that was developed for their initial study as referenced in #12? Please clarify.
6. Lines 57-59: Please clarify what “ICES” as an acronym stands for now if it was “formerly known as …”
7. Lines 59-61: Could authors provide a complete definition of their case finding in addition to the ICD-10’s used and tie this definition back to conducting a “cohort study”?
8. Lines 59-61: Could authors provide a table showing totals of the sample included in each year of the study time period in an Appendix?
9. Lines 59-61 & Table A1: “COD”?
10. Lines 59-61 & Table A1: Please explain if there were any revisions of any of those codes during the study time period. If so, what procedures were used to reconcile these revisions in this study?
11. Lines 69-74: Please provide a concise definition of how each of these databases contributed to the information contained in the database developed for this study.
12. Lines 69-74: Please provide a section explaining the quality checks that are conducted on these databases and the findings reported.
13. Lines 77-80: Again, please provide a complete definition of case finding and reference this definition here so that the definitions of “survival”, “date of diagnosis” and “most recent cancer diagnosis” will be clear to the readership.
14. Lines 84-104: Please provide short descriptions of the operationalization of these indicators in an Appendix.
15. Lines 111-115: Could authors explain how the statistical analysis conducted accounted for the time span of the data used in this study?
16. Line 117 & Table 1: Please reconcile sample sizes over here, Table 1 & Abstract.
17. Lines 117-123 & Table 1: What is the units for “survival mean”.
18. Lines 117-123 & Table 1: Are there any changes in characteristics such as “survival mean” and “age at death” over the study time period given the sex breakdown?
19. Lines 158-161: Again, could authors explain how the statistical analysis conducted accounted for the time period used in this study?
20. Lines 184-185: Again, “secondary analysis”?
21. Lines 188-190: Could authors discuss what were their results in terms of the association between sex and the “odds of experiencing aggressive care at EOL” in their initial study? Why was there a need to have a separate study on “sex-based quality of EOL care”?
22. Lines 232-233: Please reconcile with Abstract. Please see above related point.
23. Lines 248-250: Could authors describe which “outcomes” are being referenced here?
24. Please avoid typos and ensure completeness in the manuscript: a) please ensure that the references follow an appropriate format (see, for example, reference #s 5, 8 or 21, what type of a publication are those; what is the publication date for #21; why the capitalization in #25…), and b) please ensure that all acronyms are written out the first time they are reported (see, for example, “CIHI”, line 280).

Author Response

Thank you for your thoughtful and detailed review. Please see our response to the points raised below:

1. Lines 30-54: Could authors add here to explain which studies fall under what they describe as “disparities according to sex have been previously described…” in the Abstract? What were the gaps they identified in those studies? How does the current study fill those gaps?

Thank you for identifying this – we have added details accordingly.

2. Lines 56-115: Could authors add a Settings section explaining the healthcare framework within which their initial and present “population-based, retrospective cohort stud(dies)…” were performed?
3. Lines 56-115: Given that the study time horizon spans more than a decade, please explain the changes in health care services/public policy and approaches regarding EOL care and the approaches to the treatment for GI malignancies that might have an impact on the ‘quality of EOL care’.

Thank you for noting the importance of setting, service delivery, and policy. A full review of this is beyond the scope of this paper, particularly given constraints with regards to word count. An overview was provided, as well as a citation with further, more detailed information should readers require more information.

4. Lines 56-115: Please consider providing the information contained in this section broken down into sub-sections such as setting, data sources, study population… ethical considerations…

Thank you for this suggestion – sub-sections have been added.

5. Lines 56-57: Could authors explain what they mean by “secondary analysis”? Are authors using the same database that was developed for their initial study as referenced in #12? Please clarify.

This has been clarified to be from the same dataset.

6. Lines 57-59: Please clarify what “ICES” as an acronym stands for now if it was “formerly known as …”

ICES is now the official name of the organization, whereas it was formerly an acronym.

7. Lines 59-61: Could authors provide a complete definition of their case finding in addition to the ICD-10’s used and tie this definition back to conducting a “cohort study”?

Thank you for the opportunity to clarify this. We have expanded upon the creation of our dataset with the addition of Table A2, as well as clarification of how decedents were identified.

8. Lines 59-61: Could authors provide a table showing totals of the sample included in each year of the study time period in an Appendix?

Thank you for this suggestion. We have added this information in Table B1.

9. Lines 59-61 & Table A1: “COD”?

This has been revised for ‘cause of death’ to be written out in full.

10. Lines 59-61 & Table A1: Please explain if there were any revisions of any of those codes during the study time period. If so, what procedures were used to reconcile these revisions in this study?

The ICD-10 codes were not revised and were consistent through the study period.  

11. Lines 69-74: Please provide a concise definition of how each of these databases contributed to the information contained in the database developed for this study.

Please see our response to point #7.

12. Lines 69-74: Please provide a section explaining the quality checks that are conducted on these databases and the findings reported.

We agree that data quality is an important aspect of study design. We have provided a citation for how data assurance and quality is monitored by ICES. Further, these administrative data are used for billing purposes within the provincial healthcare system, and are thus cleaned and checked for quality assurance regularly by the government regulators. They are also used often for research purposes.

13. Lines 77-80: Again, please provide a complete definition of case finding and reference this definition here so that the definitions of “survival”, “date of diagnosis” and “most recent cancer diagnosis” will be clear to the readership.

As above, we have expanded upon the information provided regarding case finding. We are confident that the readers of Current Oncology will be familiar with such terms as survival and date of diagnosis.

14. Lines 84-104: Please provide short descriptions of the operationalization of these indicators in an Appendix.

Please see our response to point #7.

14. Lines 111-115: Could authors explain how the statistical analysis conducted accounted for the time span of the data used in this study?

Our statistical analysis did not account for the time span of data used in this study, with the exception of the time series graph of quality indicators (Figure 1). We felt that the inclusion of time in the analysis would unnecessarily complicate the analysis and interpretation and was beyond the scope of this paper. However, we appreciate the identification of this as a limitation in our paper and have accordingly added a line to our discussion, as follows: “Further, our multivariate analysis did not account for the time span of the data used.”

16. Line 117 & Table 1: Please reconcile sample sizes over here, Table 1 & Abstract.

This has been corrected – we very appreciate this typo being noted prior to publication.

17. Lines 117-123 & Table 1: What is the units for “survival mean”.

We have specified that the ‘survival mean’ is expressed in months.

18. Lines 117-123 & Table 1: Are there any changes in characteristics such as “survival mean” and “age at death” over the study time period given the sex breakdown?

As noted above, we did not account for time in this analysis, with the exception of the Figure 1 (run series of quality indicator proportions by sex and year of death). This is a limitation and outside the scope of the paper, which we have now identified as a limitation.

19. Lines 158-161: Again, could authors explain how the statistical analysis conducted accounted for the time period used in this study?

Please see our response to point #18.

20. Lines 184-185: Again, “secondary analysis”?

Please see our response to point #5.

21. Lines 188-190: Could authors discuss what were their results in terms of the association between sex and the “odds of experiencing aggressive care at EOL” in their initial study? Why was there a need to have a separate study on “sex-based quality of EOL care”?

Thank you for identifying this deficit, we have added a line to clarify that sex was not included as a predictor in this model. Our introduction further elaborates on the need for a separate study on how sex may impact quality of EOL care.

22. Lines 232-233: Please reconcile with Abstract. Please see above related point.

Thank you for making note of this. This line has been removed.

23. Lines 248-250: Could authors describe which “outcomes” are being referenced here?

Thank you for identifying how this sentence may cause confusion. We have clarified this to simply read, “This secondary analysis of quality indicators of EOL care over 13 years in a Canadian province provides novel and useful information regarding how sex may influence quality of care for patients.”

24. Please avoid typos and ensure completeness in the manuscript: a) please ensure that the references follow an appropriate format (see, for example, reference #s 5, 8 or 21, what type of a publication are those; what is the publication date for #21; why the capitalization in #25…), and b) please ensure that all acronyms are written out the first time they are reported (see, for example, “CIHI”, line 280).

Thank you for identifying these deficits. We believe we have corrected these by using the MDPI BibTex file for Zotero.

Reviewer 3 Report

Comments and Suggestions for Authors

Thank you for the opportunity to read this timely and relevant work.

Line 42 – The sentence describing palliative care and oncology’s relationship should be rephrased. As it reads right now, patients are diagnosed with cancer while in palliative care.

Please define palliative care. In the US, where some of your readers are, palliative care is not the same as hospice and end of life care. Hospice and palliative care are two very different types of care. I would suggest either emphasizing that the study was conducted in Canada or highlight that palliative care is for patients at EoL. So for American readers who do not realize (we American's tend to be egocentric) they are reading a study from Canada, they will not understand palliative care and EOL care are for the same patient.

Please bring in recent literature on how families benefit from what you call “aggressive care”. Although the care and treatment that a patient receives at the end of life may not prolong life or provide relief for the patient, it does serve a purpose and there is a relationship between aggressive care and bereavement. The issue this manuscript is addressing is not that patients receive aggressive, expensive, and non-curative care at EOL but that the purpose of this care is not fully understood by either the clinicians prescribing the care or the families agreeing to the care. While an extra round of chemo has no chance of curing the patient, it does provide peace of mind to families that they tried everything. The ICU or ED visits may have to do with caregiver fatigue and burnout. While this was a well-written manuscript with important results, caution must be used when using terms like “inappropriately aggressive”.

Author Response

Thank you very much for your detailed and thoughtful review. Please see our responses below:

1. Line 42 – The sentence describing palliative care and oncology’s relationship should be rephrased. As it reads right now, patients are diagnosed with cancer while in palliative care.

Thank you for noting this. We have rephrased this to read, “Within palliative care, in cases where patients are diagnosed with advanced and incurable cancer . . .”

2. Please define palliative care. In the US, where some of your readers are, palliative care is not the same as hospice and end of life care. Hospice and palliative care are two very different types of care. I would suggest either emphasizing that the study was conducted in Canada or highlight that palliative care is for patients at EoL. So for American readers who do not realize (we American's tend to be egocentric) they are reading a study from Canada, they will not understand palliative care and EOL care are for the same patient.

Thank you for this suggestion. We have added, “In the Canadian context, palliative care refers to the care provided to patients with a life-limiting illness, where the focus is upon symptom management and end-of-life care.”

3. Please bring in recent literature on how families benefit from what you call “aggressive care”. Although the care and treatment that a patient receives at the end of life may not prolong life or provide relief for the patient, it does serve a purpose and there is a relationship between aggressive care and bereavement. The issue this manuscript is addressing is not that patients receive aggressive, expensive, and non-curative care at EOL but that the purpose of this care is not fully understood by either the clinicians prescribing the care or the families agreeing to the care. While an extra round of chemo has no chance of curing the patient, it does provide peace of mind to families that they tried everything. The ICU or ED visits may have to do with caregiver fatigue and burnout. While this was a well-written manuscript with important results, caution must be used when using terms like “inappropriately aggressive”.

We agree that this is important to acknowledge. We have rephrased the sentence in question to “potentially unwanted aggressive care,” acknowledging that we cannot make such a blanket statement. We also updated this section to include a reference to Ham et al., who conducted a 2023 systematic review investigating how aggressive EOL care may impact the well-being of relatives in the bereavement period.

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

Thanks for an improved manuscript.

The following points need further attention. Old numbering is used for convenience.

3. Old “Lines 56-115: Given that the study time horizon spans more than a decade, please explain the changes in health services/public policy and approaches regarding EOL care and the approaches to the treatment for GI malignancies that might have an impact on the ‘quality of EOL care’.”

Thanks for authors’ internal response. However, this reviewer thinks that a concise description of changes over that span of more than a decade should be an essential part of this study.

7. Old “Lines 59-61: Could authors provide a complete definition of their case finding in addition to the ICD-10’s used and tie this definition back to conducting a “cohort study”?”

Thanks for adding Table A2. Please make sure that each table is self-explanatory.

15. Old “Lines 111-115: Could authors explain how the statistical analysis conducted accounted for the time horizon of the data used in this study?”

Thanks for authors’ response. However, this is such a long time horizon that not accounting for time in the analysis continues to be a concern for this reviewer (and cannot be labeled as “beyond the scope of this paper” since it is coming from the dataset that is used in this study). See a similar concern in review point #3 above.
 
18. Old “Lines 117-123 &Table 1: Are there any changes in characteristics such as “survival mean” and “age at death” over the study time horizon given the sex breakdown?”

Lack of consideration for time continues to be a concern.

19. Old “Lines 158-161: Again, could authors explain how the statistical analysis conducted accounted for the time horizon used in this study?”

21. Old “Lines 188-190: Could authors discuss what were their results in terms of the association between sex and the “odds of experiencing aggressive care at EOL” in their initial study? Why was there a need to have a separate study on “sex-based quality of EOL care”?”

Thanks. “Sex” is one of the most fundamental variables that almost all studies use. This reviewer is finding it hard to understand why it was not included as a variable in the initial analysis. Further, it is also not clear how the omission of this fundamental variable has impacted the results.

Author Response

Thank you for your thoughtful review and further feedback. Our responses are below, in italics. For simplicity, we have responded to points 15-19 together, as they concern the same issue regarding the time span of the study.

3. Old “Lines 56-115: Given that the study time horizon spans more than a decade, please explain the changes in health services/public policy and approaches regarding EOL care and the approaches to the treatment for GI malignancies that might have an impact on the ‘quality of EOL care’.”
   
   Thanks for authors’ internal response. However, this reviewer thinks that a concise description of changes over that span of more than a decade should be an essential part of this study.

We appreciate this feedback. We have added 2.1 to explain the healthcare context, which also describes some of the changes made to palliative care services in the province over the study period. This also includes references that give a more in-depth description of changes.

7. Old “Lines 59-61: Could authors provide a complete definition of their case finding in addition to the ICD-10’s used and tie this definition back to conducting a “cohort study”?”
   
   Thanks for adding Table A2. Please make sure that each table is self-explanatory.

Thank you. We have added a title to Table A2 to ensure it is clear.

15. Old “Lines 111-115: Could authors explain how the statistical analysis conducted accounted for the time horizon of the data used in this study?”
    
    Thanks for authors’ response. However, this is such a long time horizon that not accounting for time in the analysis continues to be a concern for this reviewer (and cannot be labeled as “beyond the scope of this paper” since it is coming from the dataset that is used in this study). See a similar concern in review point #3 above.
    
    18. Old “Lines 117-123 &Table 1: Are there any changes in characteristics such as “survival mean” and “age at death” over the study time horizon given the sex breakdown?”
    
    Lack of consideration for time continues to be a concern.
    
    19. Old “Lines 158-161: Again, could authors explain how the statistical analysis conducted accounted for the time horizon used in this study?”

Thank you so much for these comments about reviewing changes over time. We took your suggestions and reviewed survival over time. There were changes noted, namely that survival seems to be slightly shorter in the later years of our study period (please see the Table 1 and Table 2 in the attached document). This could be because of later diagnosis, though we don’t have enough granular data to truly speculate. However, understanding why survival changes was not the primary purpose of our paper, nor was our analysis designed for this; instead, our focus is understanding quality of care at the end of life, and specifically within the last month of life. Regardless of length of survival, because this is a decedent cohort analysis, they all have a last month of life—which we believe makes our analyses comparable. We have however, based on your comments, now incorporated time into the regression analysis. We chose to segment the study period into three time periods for ease of interpretability.

21. Old “Lines 188-190: Could authors discuss what were their results in terms of the association between sex and the “odds of experiencing aggressive care at EOL” in their initial study? Why was there a need to have a separate study on “sex-based quality of EOL care”?”
    
    Thanks. “Sex” is one of the most fundamental variables that almost all studies use. This reviewer is finding it hard to understand why it was not included as a variable in the initial analysis. Further, it is also not clear how the omission of this fundamental variable has impacted the results.

Thank you for the suggestion to include sex in the previous analysis. We did this, and it did show that there was a significant difference, with male sex associated with 27.8% higher odds of experiencing indicators of aggressive care at end of life (see Tables 3 and 4 in the attached document). However, it also showed that the rest of the findings in that paper did not change in any significant way. Therefore, this follow-up paper focussed on a sex-stratified analysis. We felt this analysis would be particularly salient given previous studies demonstrating differences by sex in goals of care, awareness of palliative care, and understanding of prognosis, as described in lines 231-246 of the discussion.

Author Response File: Author Response.pdf

Round 3

Reviewer 2 Report

Comments and Suggestions for Authors

Thanks for a revised manuscript.

Thanks for incorporating 3 time periods into the analysis. However, it needs to be clear to the readership why those 3 time periods were chosen. In the manuscript, please consider discussing the rationale clearly so that readership would understand why, first, time periods were chosen as a way to incorporate time into the analysis in general and, second, why those specific time periods were chosen.

One other minor point is:

7. “Old “Lines 59-61: Could authors provide a complete definition of their case finding in addition to the ICD-10’s used and tie this definition back to conducting a “cohort study”?”

Thanks for adding Table A2. Please make sure that each table is self-explanatory.”

Thanks. “OCR”? "RPDB”? “OHIP”…. Some of these were written out in the body of the manuscript, but all need be added here again so that the table is self-explanatory.

Author Response

1. Thanks for incorporating 3 time periods into the analysis. However, it needs to be clear to the readership why those 3 time periods were chosen. In the manuscript, please consider discussing the rationale clearly so that readership would understand why, first, time periods were chosen as a way to incorporate time into the analysis in general and, second, why those specific time periods were chosen.
   
   Thank you for this suggestion. Please see lines 140-141 for elaboration on this for the readership. 
   
   2. “Old “Lines 59-61: Could authors provide a complete definition of their case finding in addition to the ICD-10’s used and tie this definition back to conducting a “cohort study”?”
   
   Thanks for adding Table A2. Please make sure that each table is self-explanatory.”
   
   Thanks. “OCR”? "RPDB”? “OHIP”…. Some of these were written out in the body of the manuscript, but all need be added here again so that the table is self-explanatory.
   
   Thank you. Please see our revised table with a legend.