Thank you for considering our manuscript “Sex differences in mortality among a bi-national cohort of people with chronic kidney disease: do women always live longer?”, in the BMJ and the subsequent feedback to improve our study.

Detailed comments from the meeting.1. The topic of sex differences in the outcomes of an important health intervention is interesting and important for our readers and we liked that you have national data. This has the potential to be a useful paper that we would like to publish but there are two main obstacles that prevent that at the moment. Firstly the data are old and secondly the different types of renal replacement therapy are combined. If you can address these two major points then we would be pleased to see a revision of your manuscript. If that is not possible then I'm afraid we cannot offer publication in the BMJ.

Response: There is commonly a 2-3 year lag when linking datasets to death register information as registered deaths are being recorded and causes of deaths are coded into ICD. This is the case in Australia and New Zealand, as well as many other countries. In addition, data linkage studies are logistically complex, taking several months for ethics approvals, data release, cleaning and independent linkage, to the eventual release of de-identified data to researchers. The linkage exercise that formed the basis of our initial submission took 18 months to complete, added to the 3 year-data lag. This means we cannot immediately update our dataset to the most recent causes of death (which has the greatest lag-time). However, we can update fact of death, via another route, and have done so. There is substantial agreement on fact of death between ANZDATA and national death registers, where 96% of documented death dates were within 1 week of each other (Sypek M et al, Nephrology 2019). We thus used ANZDATA to update our study to the end of 2019, including updating deaths that occurred in our original study cohort until the end of 2019 and inclusion of additional patients initiating kidney replacement therapy during 2014-2019 in Australia and 2013-2019 in New Zealand. Our revised paper has updated the all-cause death analyses which was the main focus of our work, including sex differences in excess deaths (overall, by age, and by calendar year), relative survival and excess mortality ratio (EMR), and years of life lost derived from life expectancy estimates. These estimates strongly align with those previously reported in our original paper. We still provide causes of death from the national death register in Australia (1980-2013) and in New Zealand (1988-2012) as previously reported.

Our response to the comment about dialysis modalities is more nuanced. We strongly believe our work should reflect the lived experience of people with kidney failure. An individual with kidney failure commonly experiences several types of treatment in their kidney failure “career”, with periods of different types of dialysis and transplantation, often serially. While patients may begin on PD or HD, or in some instances receive a pre-emptive kidney transplant, they quite often move between all three of these therapies through the course of their life, often multiple times. Much research on this patient population does not reflect this lived experience, but is designed to align with clinician interests, or statistical ease – separation of peritoneal dialysis from haemodialysis cohorts is wholly artificial, given they are often the same people flipping treatments through time. Thus, in our initial submission we had combined all kidney replacement therapies to better reflect the lived experience of a patient receiving care for kidney failure over their lifetime, and to remove effects of selection biases that might arise by segregating patients into a dialysis type. Capturing the full effect of changing between dialysis modalities does require complex and specialized methodology (Kasza J et al, American Journal of Epidemiology 2015), and is outside of the scope of this paper but is a planned for our future work. However, we do understand presenting results by dialysis modality may aid in readers understanding the robustness of our findings. Therefore, we have provided excess deaths, relative survival and years of life lost by initial dialysis modality (PD or HD) as supplemental material. These findings strongly align with the main results, with only minor

differences. We hope that you now appreciate the reasons behind our approach and apologise this wasn’t clearer in our initial submission.

2. Please update your data to the latest year available. The most recent years are already at least 7 and 8 years in the past and this is too long. Response: As previously outlined, we have updated our study to the end of 2019 for fact of death to allow us to update the all-cause death analyses. Our updated estimates until 2019 strongly align with those reported in our original submitted paper, highlighting the robustness of our findings. Overall excess deaths remain higher among females at 11 times (SMR:11.3, 95%CI:11.2-11.5) and males at 7 times (SMR:6.9, 95%CI:6.8-6.9) the expected deaths (Figure 3). The greatest sex disparities were evident among younger females (sex interaction: p<0.01), where in those aged 18-34 years females had 54 times (SMR:53.6, 95%CI:49.2-58.4) and males had 17 times (SMR:16.7, 95%CI:15.4-18.2) the deaths expected. Trends in excess deaths over time reduced and indicated that the sex disparity has stagnated in recent years, reducing in females from 17.0 (95%CI:14.5-19.9) in 1988 to 9.2 (95%CI: 8.5-9.9) in 2010 and in males from 8.1 (95%CI:7.0-9.4) in 1988 to 6.0 (95%CI: 5.6-6.3) in 2010, but with no further improvement thereafter.

Our updated relative survival estimates, comparing the survival in the KF population to the expected survival from age-, sex-, country- and year-matched general population, are largely unchanged (Figure 4). Females with KF had a lower proportion surviving, reaching 39% (95%CI: 38-39%) at 10 years compared to males with KF who had 41% survival (95%CI: 40-42%). At any given time, females with KF had adjusted excess mortality 11% higher than males with KF (Adjusted EMR 1.11, 95%CI: 1.08-1.13, p<0.01). Consistent with previous findings, the sex difference in relative survival was not seen in those who received a kidney transplant (adjusted EMR:1.01, 95%CI:0.95-1.06, p=0.83).

The average years of life lost (YLL) remained higher in females with KF compared to males (Figure 5; Appendix Table 4). On average, females lost nearly 4 YLL (3.6 years, 95% CI: 3.6-3.7 years) more than their male counterparts. This was greater among younger females who had 5-6 YLL more than males. This sex difference was reduced in those who received a kidney transplant and was further reduced with updated data to 2019. On average, females had 2.3 YLL (95%CI:2.2-2.3 years) more than males with kidney transplant which increased to 3-4 YLL in younger ages.

We have updated these findings in the relevant results and methods section.

3. The major treatment options for renal replacement therapy are very different. Please provide separate analyses for each as well as all therapies combines as you do now. Response: As above, we had combined all kidney replacement therapies to reflect the ‘lived experience’ of a patient’s journey through care for kidney failure, which often involves all three therapies (PD, HD and transplant). Modelling the full effect of changing between dialysis modalities is difficult and requires complex methodology that is beyond the scope of this paper, and not pertinent to our research question here, but planned for our future work. Regardless, we understand that presenting estimates by dialysis modality will benefit in the interpretation of our work. Thus, we have now given the estimates of excess deaths, relative survival and years of life lost by initial dialysis modality (HD or PD) as a supplemental file. This has shown no major difference to the main findings.

Briefly, trends in excess deaths by age and over time were consistent with the main analysis (Appendix Figure 1). Younger females had the greatest burden of excess deaths and excess deaths decreased over time, but were stable since 2010. When relative survival was stratified by initial dialysis modality, females initiating with haemodialysis had excess mortality of 12% higher than

males (adjusted EMR: 1.12, 95%CI: 1.09-1.15, p<0.01). In those initiating peritoneal dialysis, the sex difference in excess mortality significantly varied by age where females aged <65 years did not have excess mortality compared to males (p>0.1). However, females aged 65-74 years had 13% excess mortality (adjusted EMR:1.13, 95%CI:1.06-1.21, p<0.01) and females aged ≥75 years had 28% excess mortality (adjusted EMR:1.28, 95%CI:1.14-1.43, p<0.01) compared to males. Females continued to lose more YLL than their male counterparts, slightly increased among those initially receiving haemodialysis (Appendix Table 5; Appendix Figure 2). Females lost an average 3.7 YLL (95% CI: 3.7-3.8 years) in haemodialysis and 3.3 YLL (95%CI: 3.2-3.3) in peritoneal dialysis.

4. In table 1 the "other" category for cause of kidney failure is rather large. Are there a few more diagnoses that could be reported separately? We wondered about other autoimmune disease, such as SLE, in women. Response: We have included further details on the other causes of kidney failure. 25% were due to “Uncertain diagnosis” where no additional clinical information was provided. This is now included as a separate category in the causes of kidney failure. We further included a footnote for the remaining 19% of the study cohort with other cause of kidney failure. The most common other cause of kidney failure was congenital abnormalities of the kidney and urinary tract (25% of this “other” category), followed by drug and heavy metal toxicity (28%), obstructions (10%), and cancer (4%). SLE and other autoimmune disease are already included in the category: “Glomerulonephritis/IgA nephropathy”.

5. We noted the higher proportion of Aboriginal women in table 1. Might this be worth drawing out a little more in the discussion? Response: We have revised the discussion to note the higher female proportion of Aboriginal and Torres Strait Islanders. This represents <10% of the KF population overall but may, to a small degree, contribute to the excess mortality among females with KF given their greater risk of death and poorer access to healthcare.

6. Please write a PPI declaration in your own words and if there was no PPI explain what the barriers were, this should go at the end of the methods. Dissemination goes in the end matter and is about how you plan to share your paper, press release, blog, policy, social media, conference etc Response: We have revised our PPI declaration. Our study was supported by a leading national patient advocacy group, Kidney Health Australia. We have recently produced an impact summary report including a lay person summary of our study outputs, such as publications, conference abstracts and prizes. We have disseminated this with Kidney Health Australia for sharing among their wider patient community, and it is freely available on our Collaborative Centre Organ Donation Evidence website (https://organdonationevidence.org.au/category/researchoutput/). We used existing data collected in national registers in Australia and New Zealand. Due to the sensitive nature of the data collected, and our ethics governance permissions, we will not be sharing the raw or linked data for public use.

7. Please revise your paper to respond to all of the comments by the reviewers. Their reports are available at the end of this letter, below. In your response please provide, point by point, your replies to the comments made by the reviewers and the editors, explaining how you have dealt with them in the paper.Response: Please find our point by point responses described.

Reviewer: 1Thank you for the opportunity to review this very interesting paper. This is a lay review. I am a CKD patient (consequent to AKI due to TSS) and this review is written through that lens. As patient reviewers we are given guide questions which I will use to shape this review.In evaluating sex differences in mortality in people with KF compared to the general population the

authors would seem to be addressing an extremely relevant and important, indeed fundamental, question, especially if, as would seem to be the case from their results, the answer to this question is different from that gained by simply comparing the mortality in the different sexes within the KF population. It is for others to assess the robustness of the methods and statistics of this paper, but if the conclusions are sound the findings of this research would have great significance for patients as, as the authors suggest, it would raise the question of why it is the case that 'the life limiting impact of KF is more profound for women'(p10), and what can be done about it. The authors make some initial suggestions as to some reasons for these sex differences. This research could be the launchpad for further research which could lead to potential improvements in the lives of individual patients. The question and the outcome therefore are of relevance and significance in a very direct and practical way.Response: Thank you very much for kind review.

The authors have set out to answer a specific question and have described their findings - they have pointed to further questions which result from it. The conduct of their research and any outcomes from it have not caused any difficulties or challenges for patients as the authors have used available data and there was/is no need for intervention with patients at this stage. Perhaps this is why they felt there was no need for patient involvement in the research? I do have some concerns about this. Patient groups could have been involved in discussing the research question, for example, and the implications for patients. Such a dialogue might have formed a basis for future joint working on the questions which result from the study.Response: Apologies for our lack of understanding. Our study did not involve ‘research buddies’ such as patient representatives of those with KF. The research question was not developed with patient involvement, but instead was based on research priorities set by patients and caregivers with chronic kidney disease where mortality ranked in the top 10 priorities (Urquhart-Secord R et al, American Journal of Kidney Disease 2016; Morton R et al, Nephrology Dialysis Transplantation 2011).

In addition, our study was supported by a leading national patient advocacy group, Kidney Health Australia via a competitive funding process. We have recently produced an impact report that provides lay-person summaries of our study findings and outputs (eg. publications, conference abstracts, prizes). This report was shared with Kidney Health Australia for disseminating with their wider patient community and audience, and will also be publicly available on our research group website, the Collaborative Centre for Organ Donation Evidence (https://organdonationevidence.org.au/category/researchoutput/).

I am particularly puzzled by the statement that it is not possible or appropriate to disseminate the results to patient organisations. Further explanation of this from the authors would be welcome. As I have suggested the findings that 'the female survival advantage seen in the general population is entirely lost in the presence of KF'(p3) and the suggestions that this may be due to 'inequities in health service deliveries' and 'systemic differences in care' (p4) are very relevant to patients and I am not clear as to why it is not appropriate to share the findings with them. It is, so to speak, my body and my disease and I am uncomfortable knowing that there are academics and clinicians who think I shouldn't be told. Indeed this stands in contrast to the authors saying that their 'findings will inform both clinicians and patients with KF about expected survival outcomes' (p13). The authors therefore do seem to wish their findings to be shared with patients, and I would have thought it possible and indeed appropriate to share this research with patient organisations, including the researchers' funder. Response: Thank you so much for reminding us about the importance of patient involvement. We apologize for our lack of clarity. We used existing data collected in national registers in Australia and New Zealand, so we were unable to directly inform patients who had contributed data of our research due to existing privacy protection and consent agreements. However, when starting

treatment for kidney failure in ANZ, patients are informed about the ANZDATA register and are given the opportunity to opt out. In the last 15 years this option has not been pursued by any patients. There are extensive patient resources available about this on the ANZDATA website anzdata.org.au

As mentioned above, our research was supported by Kidney Health Australia, a leading patient advocacy group, where we have disseminated lay-person summaries for sharing among the wider community. We are committed to disseminating our research findings with patients with KF, caregivers and the community. We strive to continue to produce summary reports with lay-person summaries to be made publicly available via patient advocacy groups, such as Kidney Health Australia, as well as our research group website.

The paper is clear and easy to read as it stands but it would be possible to create a 'lay friendly' version in order to disseminate the findings more widely. Without further clarification on this from the authors, as to why they do not plan to disseminate the results, there might be a suggestion of paternalism which I am sure would not be intended. It is only because I would see this research, and the questions arising from it, as being so relevant and potentially important to patients that I have these concerns. It might be good for patients and patient organisations to have the benefit of it and the opportunity to share in future research.Response: Thank you for your suggestion. We have addressed these concerns in our above responses. As you have recommended, we will also create a 1 page lay-person summary specifically for this paper which we will share with Kidney Health Australia to disseminate to the community with KF and wider audience, as well as be made publicly available on our research group CODE website. We will produce a twitter-friendly picture summary to be shared via social media to have wider reach in the community.

Reviewer: 2The manuscript by De La Mata eta al. evaluates mortality differences between patients with end stage kidney disease and general population in Australia and New Zealand, with emphasis on the sex differences. This is important and interesting work performed on a high-quality data. However, there are multiple opportunities for improvement and need for some clarifications.

1. The term “kidney failure” is rather old-fashioned, please use instead “end stage kidney disease” (ESKD).Response: The Kidney Disease Improving Global Outcomes (KDIGO) released in May 2020 a nomenclature recommending preferred terms for more precise, consistent and patient-centered description of kidney function and disease. In this release, they recommend abandoning the use of non-patient centered terms such as “end-stage” or “renal” and instead using the term “kidney” and “kidney failure”. Please see for more information: https://kdigo.org/kdigo-announces-publication-of-the-nomenclature-for-kidney-function-and-disease-conference-report-and-glossary/

Thus, we have not made any changes to the current use of “kidney failure” term in our paper.

2. One of the major shortcomings of the manuscript is the merging of all patients on kidney replacement therapy (KRT) in a single category (dialysis and kidney transplantation together). The mortality is so different (several folds) between dialysis and kidney transplantation patients that they have to be analyzed separately. The authors analyzed outcomes in the “whole KRT group”, and then separately analyzed kidney transplant patients. But analysis of “whole KRT group” is not very useful due to extreme differences in mortality between dialysis and kidney transplantation. Moreover, the dialysis is also not uniform method, and at least hemodialysis and peritoneal dialysis groups should be separated. Further, the kidney transplantation results could vary substantially between living and deceased donation that requires their separate consideration.

Thus, the analysis of the nation-wide data should include 4 categories of patients: hemodialysis, peritoneal dialysis, living kidney transplant, deceased kidney transplant – and compare them separately to the general population for calculating SMRs This will substantially increase the value of the manuscript, and provide important data.Response: As explained above, we had combined all kidney replacement therapies as most patients experience both dialysis modalities at least once so we wanted to reflect the patient’s lived experience. However, we understand presenting our findings by dialysis modality may aid in interpreting the robustness of our work. Therefore, we have now provided estimates of excess deaths, relative survival and years of life lost by initial dialysis modality (HD or PD) in the results and in the appendix. There were no major differences to the main findings. The average YLL was slightly greater among females initiating haemodialysis than peritoneal dialysis. Please refer to above response 3 for further explanation, as well as amended results and methods.

3. In the “Abstract”:a. The first two phrases of the “Results” are representing “Conclusions”, thus it would be better to rearrange the abstract.Response: We have rephrased the results section of the abstract

b. The authors have to clearly indicate whether the numbers represents ESKD on kidney replacement therapy (KRT), and better distinguish studied metrics among dialysis (preferably by dialysis modality) and kidney transplant patients.Response: We have amended our methods to note that pre-dialysis patients with KF not receiving kidney replacement therapy and people with KF receiving palliative care were not included. As previously outlined, we had combined all kidney replacement therapies to provide a ‘lived experience’ of a patient receiving care for KF, which often involves both dialysis modalities and kidney transplant, for a third of patients. However, we understand presenting our findings by initial dialysis modality may aid in interpreting our findings and their robustness. As such, we have provided findings stratified by dialysis modality for excess deaths, excess mortality and years of life lost (YLL). There were no major differences to our original findings.

4. In the summary bullet points and elsewhere some items should be expressed more clear and straight. For example, “The life-limiting impact of KF is …” could be changed with smth like “Mortality among patients with ESKD is …”. Again, the difference between dialysis (preferably by dialysis modality) and kidney transplant patients should be better distinguished.Response: Please refer to above response. We have now presenting findings by dialysis modality.

5. In the “Introduction”:a. First the literature epidemiologic data are provided, then the CKD definition is given, and after another epidemiologic data are provided. It would be much more logic to give all epidemiologic data together.Response: We have revised the introduction paragraph to initially provide a definition of CKD, then describe the epidemiology of CKD and burden on the healthcare system and individuals.

b. Please clarify “once receiving dialysis females have lower rates of cardiac catherization andcardiac rehabilitation”. In general, the “Introduction” could be reviewed and restructured to provide more clear intro supported by major numbers of literature references available for the sex differences in CKD epidemiology.Response: We have made revisions to the introduction to improve clarity as well as provided more relevant references.

6. Some phrases should be changed for clarity and easy to read. For example, “However, inequities

in health service delivery for women means the female survival advantage may not always persist.” Some phrases are rather clear but should be reviewed for the style, for example “Females are also underrepresented in randomized clinical trials and animal models, which can lead to …”. Some phrases, in addition to revision, need support by references, for example “To some extent this is mitigated by the use of eGFR to monitor changes in kidney function, which accounts for sex, but women may still be referred for specialist care later than men with comparable kidney function.”Response: We have revised the suggested sentences for clarity and easy of reading and included a reference for the given sentence.

7. In the “Methods”:a. The authors clearly indicated that only dialysis and kidney transplant patients were included, and this is important to note in the “Abstract” and in the “Discussion” that persons with pre-dialysis ESKD not on RRT and persons with ESKD on palliative care were not included.Response: We have amended the methods to note that pre-dialysis people with KF not receiving kidney replacement therapy or people with KF receiving palliative care were not included. We did not amend the abstract as we were limited by the word count.

b. It should be described how the Australian Modification is different from “classical”ICD-10, and whether these differences are relevant to the performed analysis.Response: There are no differences in the ICD-10-AM and ICD-10 for cause of death information. Only minor differences exist in ICD-10-AM for diagnoses relating to hospital admissions, but this is not relevant to this paper. For simplicity, we have removed all reference to ICD-10-AM and replaced it with ICD-10.

c. The authors performed an excellent work by including patients started RRT in 1988/90-2012/2013. Given the long period, the time trends should be reported.Response: Please refer to Figure 3C where we compare excess deaths over time. While initially excess deaths were decreasing more rapidly in females with KF, this has stagnated in recent years with a persistent sex disparity. Females with KF continue to experience a greater burden of excess deaths compared to their male counterparts. We have revised the discussion to mention this and express the need for further research that could address this sex disparity.

d. The reasons why the analysis is limited to only 2012/2013 and not cover the most recent years (which could be different in the studied metrics) should be provided. Preferably, the analysis should be extended to the contemporary years.Response: As discussed above, it is common to have a lag of 2-3 years when linking to national death registers as registered deaths are being recorded and causes of deaths are clinically coded into ICD. However, we understand the need to report recent trends and have now updated the study cohort and follow-up until the end of 2019. We determined fact of death from ANZDATA during 2013-2019 for Australia and 2012-2019 for New Zealand. Our other study has shown that fact of death is in strong agreement between ANZDATA and national death registers, where 96% of death dates occur within 1 week of each other. There are no major changes to our reported findings using the updated cohort. Importantly, trends in recent years has shown that the sex disparity has not converged and remains constant since 2010.

e. The authors indicated “Australian patients with additional records after the registered date of death were considered to be alive and censored ”. The percentage of such patients should be reported.Response: We have now included that 17 patients (<0.1%) were censored at the registered date of death as they had additional records in ANZDATA.

f. The authors stated “Percentages for the causes of death were compared between sexes using Chi-squared tests to examine trends across age groups and linear regression to examine trends over time ”. However, the calculation of mortality rates per 100 patient-years and use of Poisson regression with comparison of mortality rates is much more adequate for this task, with evaluating p for trend to estimate the differences over time.Response: We agree linear regression is not ideal to compare the proportions attributed to the different causes of death. A Poisson regression would be appropriate for comparing rates, but here we compare proportions. We have changed the statistical method to use logistic regression instead. We no longer found interactions between sex and age (p>0.001), however all other findings were relatively similar. Across ages, males had higher proportion of cancer deaths (3.3%, 95%CI: 2.6-3.9%) and cardiovascular deaths (5.0%, 95%CI: 4.0-5.9%) compared to females. While females had higher proportion of deaths attributed to chronic kidney disease (5.2%, 95%CI: 4.3-6.2%) and gastrointestinal disease (0.9%, 95%CI: 0.5-1.3%). Similar trends were seen over time: males had higher cancer deaths (3.1%, 95%CI: 2.4-3.7%) and cardiovascular deaths (5.3%, 95%CI: 4.4-6.2%), females had higher deaths from chronic kidney disease (4.6%, 95%CI: 3.7-5.5%) and gastrointestinal disease (0.9%, 95%CI: 0.5-1.4%). There was no sex difference over time or by sex for deaths attributed to respiratory disease or diabetes (p>0.1).

g. In the part “5 year age band, sex and calendar year. These data are publicly available from the Australian Bureau of Statistics and the New Zealand Ministry of Health. ”the reference to the national bureau is required.Response: We have revised to include references to the Australian Bureau of Statistics and New Zealand Ministry of Health public summary of death data.

h. The description “The life expectancy was estimated, and thereafter the expected years life lost (YLL), …” is not sufficient. More details on estimation of life expectancy are needed.Response: Please refer to the fourth paragraph, second last sentence of “Statistical analyses” section in the methods. Here we state “Average YLL were calculated as the difference between the estimated life expectancy in our study population and life expectancy in the general population, matching on 5-year age band at initiation of kidney replacement therapy, sex, calendar year and country [21, 22].”

i. The assumption about 110 maximal life expectancy in the phrase “The time at risk was altered to be relative to the patient’s date of birth, such that the hazard of death was a function of a patient’s age until 110 years of age.” seems somewhat too optimistic. Moreover, the phrase needs clarification.Response: This survival modelling approach requires us to set an upper age limit, which is impossible for participants to exceed. Our choice of 110 years is not uncommon in these models. We had modelled the survival curve until age 110, but as life expectancy is determined from the area under the curve, it does not falsely inflate our estimates of life expectancy as the probability of survival is exceptionally low at these ages (though possible). We have removed “until 110 years of age” and added this later on when describing life expectancy as the area under the curve until age 110.

j. The authors need to report whether patients re-starting dialysis due to kidney transplant failure were included in the study, and how their at-risk periods were calculated.Response: In the “Death ascertainment and cause of death” section of the methods, we have described that time at risk was measured from date of kidney replacement therapy for the KF population and the earliest transplant date for the subpopulation of kidney transplant recipients, until date of death or censored at 31st Dec 2019 (updated). We have clarified that we considered an once transplant, always transplanted analogous to an ‘intention to treat’ approach. It is difficult to

disentangle the effects of returning to dialysis and mortality thereafter in the kidney transplant recipient population. Censoring after graft failure will not capture all deaths in this population, despite being intrinsically related to persons health and survival after kidney transplant. We believe our approach provides the best lived experience of a typical kidney transplant recipient.

k. It has to be reported whether the pre-emptive kidney transplantations were included in the analysis, and whether both deceased and living donations were analyze.Response: We have described the number of pre-emptive transplants in Table 1. These represent a minority of kidney transplant recipients overall (~10%). The kidney transplant population includes both deceased and living donors. We have clarified this in the methods.

l. In the phrase “Patients with missing data for smoking history were excluded from the model.” the percentage of such patients should be reported.Response: We have amended this sentence as smoking history was not routinely collected until after 1991. The number of patients without smoking history is reported in Table 1, with <10% missing smoking history. Therefore, we revised our methodology to undertake multiple imputation chained equations to impute smoking history for this <10% of patients with multinomial logistic regression of 5 iterations. There are now no people excluded from the model.

m. More details should be provided regarding “An additional time varying covariate was included for the first 6 months of follow-up to adjust for increased mortality after initiation of dialysis or transplantation.”Response: To increase the precision of our life expectancy estimates, we included an additional time-varying binary covariate to indicate the first 6 months of follow-up to adjust for possible differences in mortality in the initial 6 months following kidney replacement therapy. This covariate was significant in the models for kidney transplant, HD and PD. For kidney transplant and HD, the first 6 months conferred an increased mortality risk, while for PD the first 6 months reduced mortality risk compared to the remaining follow-up. We have clarified this in the methods by describing it as a binary covariate to increase precise of our predicted survival for life expectancy estimates.

n. In general, the “Methods” should be supported with more literature references to the applied methods, since many of them are named but rather briefly described in the manuscript.Response: Our methods section has referenced the statistical methodology for modelling excess mortality ratios (EMR) from relative survival [15]. The estimation of life expectancy as the area under a survival curve is a common epidemiological method, however we have now provided a reference for that. Similarly, indirect standardization is a common epidemiological approach and we have not provided a reference.

8. In the “Results”:a. The sub-section “Sex differences within the population with KF” contains a mixture of described categories expressed as two percentages in the round brackets. At the beginning the two percentages in brackets compare males to females, then the two percentages in brackets compare females to males, then the two percentages in brackets indicates different disease for the same sex. Please review this part for simplicity of description and interpretation.Response: We agree this would be confusing for readers. We have amended this results paragraph to consistently present female percentages vs male percentages.

b. At the Figure 1 the cardiovascular diseases (CVD) not exceeding 25% of all deaths even in the advanced age groups. This should be discussed and compared to the literature data from other countries.

Response: As mentioned in the methods, we reported the causes of death as given by the primary cause of death. The primary cause of death is defined as the disease or condition that initiated the sequence of events that resulted in death. It is difficult to disentangle the degree to which a disease or condition contributed to the death when relying on both the primary and secondary causes of death. However, when also looking at secondary causes of death we do find that the % of cardiovascular disease in our study cohort increases to 40% of all deaths. Our paper has not discussed this as we have another paper in submission elsewhere that evaluates cardiovascular deaths in the KF population.

c. At the Figure 2 the proportion of deaths for males due to CVD decreased by almost 10% over time, a finding that worth to discuss and propose some explanations for this.Response: We found no evidence to suggest males with KF had a significantly greater decrease in the percentage of CVD deaths over time compared to females with KF (i.e. no interaction between sex and calendar year – this was mentioned in the last sentence of the “Sex differences within the population with KF” section of the results). We have added to the results that deaths attributed to cardiovascular disease has reduced the most over time by 5.2% (95%CI: 3.2-7.2%; p<0.001) in both sexes. Previously we had mentioned the greater use of cardiovascular prevention likely benefiting the population with KF over time, we have now referred to the decreased percentage of deaths attributed to cardiovascular disease.

d. At the Figure 5 it is unclear why the x-axis is limited to 60-year old only (the appendix 4 is also limited to 60 years old only). Please provide data for all ages.Response: We have extended the life expectancy estimates to 80 overall and by initial dialysis modality, grouping all those ≥80 years as this approaches general population life expectancy estimates and those receiving kidney replacement therapy in those ages are a very selective group of people. Our findings may be misinterpreted to suggest those receiving kidney replacement therapy after 80 years have greater survival than the general population. Similarly, we have presented life expectancy estimates until age 70, grouping all those ≥70 years. There are fewer kidney transplant recipients beyond 70 years of age, and again this reflects a very select group of people. For ease of comparison, we have extended Figure 5 to age 70, footnoting that age 70 includes 70 and above for transplant population and age 15 includes less than 15 years.

9. In the “Discussion”:a. At the figure 3 the absolute mortality rates for males and females on KRT (panel A) are not different, while the SMRs (panel B) show prominent sex differences. The authors describe these differences in the “Discussion” correctly but very laconic by the single phrase “Sex differences in mortality were only apparent once compared to the general population.”. This message should be presented in more details.Response: The lack of sex differences in panel A is relevant, as females should have lower mortality rates than men to reflect general population trends. Our study aimed to evaluate sex differences in mortality using relative measures of survival to account for expected background mortality and understand the life-limiting impact of KF. The focus of our work is not to make comparisons within the population with KF, but relative to expected survival outcomes in the general population.

b. Possible explanations of the decrease in SMRs in both sexes over time are very welcome.Response: The end of the third paragraph in the discussion section explains possible reasons for the reduction in excess deaths over time, specifically more kidney transplantation and better graft survival. Cardiovascular and stroke deaths are also a driving factor as greater use of cardiovascular prevention has likely benefited the population with KF. We have also revised this section to highlight the need for further research into continuing to decrease the sex disparity in excess deaths, such as improvements to patient care for KF or better access to healthcare and services.

c. The phrase “The sex disparity in excess mortality seen in our study was greater than previous reports in cardiovascular disease [17, 18] and cancer [19].” and subsequent considerations should clearly indicate that literature data from refs 17-19 considers general population, and not ESKD. In general, the “Discussion” should better distinguish data from general population from data from ESKD population.Response: We have amended this sentence to indicate these reports were in the general population for cardiovascular disease and cancer.

d. The authors mainly considers “Reasons for comparably worse survival outcomes in females with KF can be attributed to differences in the delivery and quality of care for KF [5].” that could be one of the reasons but are not supported by their own data. The absolute mortality rates for males and females on KRT (fig 3 panel A) are not different – that indicates the equal treatment quality in Australia and NZ for both sexes. Only the SMRs are different, and this has to be interpreted appropriately. This one-sided interpretation is among major manuscript shortcomings.Response: We have amended this sentence to indicate differences in the delivery and quality of care “may” be attributing to the comparably worse survival outcomes in females with KF. The mortality rates not being different between sexes is an important finding in itself, in the general population it is expected that females have lower mortality rates than males. The basis of our work is to provide a comparison to expected outcomes in the general population. If the life-limiting impact of kidney failure was the same in females and males, then we would expect similar excess deaths and excess mortality. Instead, we find that females with KF comparably lose more survival than their male counterparts with KF. The female survival advantage no longer exists in the presence of KF and has important implications for patient communication of expected survival outcomes. Further, while excess deaths were decreasing at a faster rate in females with KF, this has stagnated in recent years with a constant sex disparity since 2010.

e. More literature data on the topic exists, including one of the first demonstrations of mortality differences by sex between general population, dialysis and kidney transplant patients in the USRDS 1997 ADR. The results of CKD-PC could also be used as a reference point in comparison of the study findings. Citation of these and other references would substantially strengthen the manuscript.Response: We agree there may be other published literature on this topic, however we tried to limit the references to the most relevant and recently published work. Excess deaths are no longer reported in USRDS Annual Data Reports, however the 2020 ADR does report expected life years remaining. We have revised our discussion to compare with their results. Our life expectancy estimates were slightly higher, however depicted similar trends where females receiving dialysis had lower life years remaining and higher years of life lost compared to their male counterparts.

f. Considering the need to analyze separately 4 categories of patients on KRT, the “Results” and “Discussion” have to be substantially changed.Response: As discussed above, patients often experience both dialysis modalities over their lifetime for care of KF and our combined analyses provides an average lived experience. However, we understand findings presented by dialysis modality may aid in interpreting our findings. Thus, we have presented excess deaths, excess mortality and years of life lost (YLL) estimates by initial dialysis modality in the appendix. There were no major changes to the main findings. We have amended our methods and results to discuss this.

10. Finally, the title could be changed to more informative one, with indication of cohort registry data use.Response: We have amended our title to highlight the use of a bi-national cohort.

Reviewer: 3The authors present an interesting population cohort linkage analysis of their renal registry up to 2013 in 60,823 patients with 57% deaths - assuming a primary hypothesis that there is no sex difference in survival with renal replacement therapy.

As a non statistician I would recommend review in addition by a statistician to ensure the full analysis is appropriate.

The literature has already shown sex differences in mortality in other populations and this current analysis adds further data to in essence confirm that end stage renal disease negates the survival benefit in females (especially the young) in the general population but transplantation does restore this to some degree.

The paper is well written and I have a few suggestions for the authors to consider;

1. The data being limited to a population in Australia and New Zealand does limit the generalisability of the data and it would be important to indicate this.Response: Our methods indicate this is a population-based study, including the all people receiving kidney replacement therapy in both Australia (1980-2019) and New Zealand (1988-2019) over nearly 40 years. This was highlighted as a major strength of our study in the discussion as it encompasses the entire population of KF, this will not be subject to selection bias and provide the least biased results. Our findings are highly generalizable to Australia and New Zealand, as well as other developed countries that offer universal healthcare, such as UK and Canada. We have amended our discussion to highlight the generalizability of our findings as a major strength of our study.

2. Do the authors think the higher % aboriginals although 10% vs 5% ( a small percentage) is significant as it is recognised this population has a higher mortality, poor compliance with RRT and poor access to health care. In addition the transplant rates tend to be lower in this population and may contribute to a higher mortality.Response: Yes, we do note that females have a higher proportion of Aboriginal and Torres Strait Islander compared to males. Although these do represent the minority of the population overall (<10%), which makes it difficult to examine this as a subpopulation and are unlikely to have a major impact on our findings. We have revised our discussion to highlight the greater proportion of Aboriginal and Torres Strait Islander in females which may contribute, to a small degree, to the excess mortality.

3. Is there any impact of lead time bias on commencing RRT - I.e. that females commence later than males, although I note the comparisons over time are similar.Response: Previous studies have shown that females do tend to initiate dialysis later than men, however a randomized, clinical trial (the IDEAL study) showed that late dialysis initiation does not impact survival (Cooper et al, The New England Journal of Medicine 2010). We have also now presented the previous dialysis duration before first transplant in Table 1, which found no differences between sexes.

4. The data do not explain why a higher mortality in younger females or have I missed this. This is one area where the paper is somewhat disappointing as it does not provide any reason for similar deaths between sexes and other outcome when RRT therapy is started.Response: The focus of our paper was to evaluate whether sex differences in mortality exist in the population with KF, relative to expected survival outcomes in the general population, to build evidence for future research to inquire about why these sex differences have arisen. Our paper provides estimates of relative measures of survival to understand excess deaths and excess

mortality, as well as years of life lost. Without our study findings, there would be no rationale to support the need to explore why younger females have a greater burden of excess mortality.

5. As the authors mention due to lag in data they have examined up to 2013 but with the improving trend in CV outcomes what do the authors speculate the data may look like in the last 5 years as this would be more relevant now compared to the historical data. The analysis over time does provide useful information but there have been a number of changes in the last 7 years such as movement to Hemodiafiltration; improved fluid balance; a better understanding of sudden cardiac death and attempt to mitigate this; other dialysis options.Response: As described above, we have updated the study analysis to include further deaths and update the study cohort until the end of 2019. While our findings were similar to previously reported, we were able to extend trends in excess deaths until 2019. Females with KF continued to have excess deaths throughout the study period. Excess deaths in females initially declined at a faster rate than males, but stagnated in recent years where there have been no improvements in reducing the sex disparity since 2010. Our updated estimates of years of life lost (YLL) have indicated reduced YLL in the transplant population, likely driven by greater access to transplantation and improved post-transplant survival.

6. Just for clarity this population included both haemodialysis and peritoneal dialysis and if I missed it what was the split.Response: Yes, our main findings present estimates for the entire population receiving kidney replacement therapy, including both haemodialysis and peritoneal dialysis. These were combined to present the lived experience for patients with KF, who often experience both dialysis modality over their lifetime. However, as recommended above, we have now presented estimates stratified by initial dialysis modality in the Appendix. There were no major differences in the study findings, where younger females continue to have a greater burden of excess mortality and more YLL. Please refer to initial response 3 for further details.

7. A factors which seem to stand out in females is there was a significant proportion underweight (this might reflect malnutrition; delayed dialysis etc) . In addition the higher mortality was related in part to gastro-intestinal causes ( a low % overall) and Cardiovascular causes. Can the authors offer any reason for this finding in their population and was it present in other studies.Response: While females had a higher proportion who were underweight compared to males, this only represented a small minority (7%). A randomized controlled trial found no evidence that delayed dialysis impacts upon mortality (Cooper et al, The New England Journal of Medicine 2010).

Females did have more excess gastrointestinal deaths, but these did represent a minority of deaths overall (4% in both males and females). To the best of our knowledge, there are no other studies in the population with KF that evaluate excess mortality for gastrointestinal causes. Previous studies have not found an increased risk of gastrointestinal diseases among women receiving dialysis, suggesting females may have poorer responses to treatment and long-term outcomes. We have revised our discussion to include this.

8. Figure 2 presents interesting comparisons of death aetiology between sexes; it might be useful to censor for transplantation, although I note figure 5 does present the transplant comparisons.Response: As discussed above and following, patients with KF often experience both dialysis modalities and kidney transplantation through their lifetime. We combined all kidney replacement therapies to present the lived experienced of a patient with KF. Capturing the full effect of changing between transplant to dialysis, and between dialysis modalities is complex. Censoring after transplant will not capture many deaths that occur after graft failure, but still intrinsically relate to a

person’s health and survival outcomes after a kidney transplant. Complex and specialized methodology are required and is outside of the scope of this paper, but is planned for future work.

Reviewer: 4Thank you for the opportunity to review this manuscript. In this population-based retrospective cohort study, De La Mata and colleagues investigated differences in mortality by sex among individuals with kidney failure, comparing the observed mortality rates to those expected in the general population in New Zealand and Australia. The findings are quite sobering. Women with kidney failure had markedly higher mortality rates compared to men, culminating in substantial years of life lost especially among younger women. This excess relative mortality has diminished over time, but the gap remains significant in the contemporary population of people with kidney failure.

I think this is a great paper and it ought to be replicated in other jurisdictions given the importance of the findings. The study design is appropriate and well-executed, the findings are thoughtfully presented, and the information regarding cause of death is very comprehensive. I have some comments and suggestions for the authors’ consideration.

Major comments1. In several places in the manuscript, the authors posit that women with kidney failure may receive suboptimal care because of factors such as delayed recognition of kidney disease and late referral to specialist care. Did the authors consider examining intermediate outcomes to support this hypothesis such as vascular access (starting dialysis with a line versus a fistula), modality (in-centre versus home therapies), pre-emptive transplantation, or duration of pre-dialysis care? If this information is not available in ANZDATA, is there evidence in the literature to support this?Response: We did not explore additional factors relating to care for KF, as mentioned by the reviewer. The focus of this paper to review sex differences in mortality overall, where we present several relative measures of survival that adjust for expected background mortality. ANZDATA has some measures relating to dialysis care, specifically line vs fistula and ‘late referral’ (measuring those presenting for the first time within 3 months of needing to start dialysis). Pre-emptive transplants only represent a minority (3%) of population with KF over nearly 40 years, as given in Table 1, making it difficult to evaluate trends in a very select group of people. We thank the reviewer for their great suggestions and plan to evaluate how dialysis measures and other factors (eg. home vs in-centre, social determinants of health such as socioeconomic status) impact upon relative survival outcomes in our further work.

2. The findings for the subpopulation of transplant recipients indicate that the survival differences are not as marked between men and women. I believe the transplant subpopulation is nested within the primary analytic cohort, as opposed to being a separate distinct sample. From the design, it seems that individuals were followed from date of dialysis until death or the end of the observation period. How was transplantation handled during follow-up? Given that it clearly changes the survival probability, should a transplant event be treated as a time-varying covariate?Response: Yes, you are correct. The kidney transplant recipients are a subpopulation that are nested within the primary analytic cohort. However, the vast majority of the transplant recipients (>90%) had received a kidney transplant after initiating dialysis. All estimates in the transplant population were from first kidney transplant until death or the end of follow-up (end of 2019), considered as once transplanted always transplanted. Capturing the full effect of changing between transplant to dialysis, and between dialysis modalities is complex and time-varying covariates do not reflect the true effect of subsequent dialysis after graft failure, nor mortality following transplant. While possible to do, we judged this was outside of the scope of this paper (Kasza J et al, American Journal of Epidemiology 2015), but we are planning to undertake a methodology study to explore this more

granularly. We believe our estimates present the lived experience of a typical kidney transplant recipient, and we have further provided estimates by initial dialysis modality, which has shown little difference to the main findings.

3. One obvious difference between men and women with respect to transplantation is the higher likelihood of prior sensitizing events in women. I see that the authors included dialysis duration as a covariate in the transplant model. Were there differences between men and women in the distribution of time from dialysis start date to transplant date? Might differences in waiting times have changed over time due to changes in transplant allocation policies?Response: Yes, previous dialysis duration was included in predicting the life expectancy and thereafter the years of life lost (YLL). We have now included in Table 1 the distribution of previous dialysis duration before first transplant. There were no significant differences in the distribution of prior dialysis duration, where both females and males had a median of approximately 20 months on dialysis before their first transplant.

4. Some of the methods/results may need some clarification:a. Figure 2 shows differences in causes of death by sex over time. The results state that males had a 3% higher proportion of deaths from cancer and 5% higher from CV disease. Where do these summary estimates come from? Is this a difference in slope from a linear regression model? If so, it would be more meaningful to provide a confidence interval for the estimate rather than a p-value. If it is a slope, is there an assumption that it is linear across time? The data for CV disease would suggest that it is not.Response: In the first paragraph of “Statistical analyses” in the methods, we stated that percentages of causes of death between sexes were compared using Chi-squared for trends across age groups and linear regression to examine trends over time. However, given the previous reviewer 2 comments, we have revised our methods to use logistic regression to compare proportions of causes of death between sexes by age group and by calendar year. We have now reported the point estimates with 95%CI of the percentage differences. There were few changes to previously reported sex difference in causes of death.

b. Where does the fitted line in Figure 3C come from? Is it a Lowess line or something similar?Response: Yes, the fitted line is a lowess line. We have included a footnote to clarify this.

c. Is there an assumption that the EMR is constant over time, analogous to a HR from a Cox regression model? If so, this assumption should be tested because the preceding mortality data suggest that sex differences in mortality have narrowed over time.Response: The referenced Dickman et al paper discusses the application of relative survival modelling. There is an assumption that excess hazards are proportional over follow-up time, analogous to the proportional hazards assumption in a Cox model. However, non-proportional excess hazards can be incorporated through the use of interaction terms between follow-up time and covariates. Testing the interaction between follow-up time and year period of initiation of kidney replacement therapy was significant (p<0.001), but it did not change our estimates of excess mortality among women with an EMR of 1.11 (95%CI: 1.08-1.13). We chose not to include any interaction terms as it complicates the interpretation of findings with little difference in EMR estimates.

d. Why are the covariates different for the Poisson model and the survival model?Response: The relative survival model was used to estimate the excess mortality ratio (EMR), where the most significant covariates likely to confound the association were included. Due to computational burden, we limited the number of covariates to still allow for robust and accurate measures of excess mortality in women compared to men. However, the parametric survival model

was used for predicting survival in our cohort to estimate the life expectancy and years of life lost thereafter. Thus, we included as many significant covariates as possible to improve prediction.

5. Has the probabilistic linkage method been validated previously?Response: Yes, as mentioned in the limitations paragraph the false positive link is low and estimated to be <5 per 1000 records. We have now added a reference for this estimate. The fact of death is also in high agreement between ANZDATA and national death registers, further supporting the accuracy and reliability of linkage.

6. The dramatic difference in mortality in younger age categories merits greater emphasis in the discussion. Similarly, reporting the average difference in YLL masks the huge differences in YLL at younger ages. Why do the authors think that this might be the case? Connective tissue diseases were one of the most common causes of death in younger women compared to their male counterparts, and they amplify the risk for cardiovascular events. Could this have contributed to the marked disparity in younger people? Might the presence of systemic conditions partly explain why young women also had greater YLL compared to young men even after transplantation (Figure 5B)?Response: We hypothesis that cardiovascular disease has a significant role in the excess deaths among young women. We have made minor edits to the paragraph in that discussion that refers the excess mortality among young females. Briefly, there is likely an under-recognition of cardiovascular risk in young women. This is supported by an Australian study in the general population (the AusHEART study) which found general practitioners are more likely to assign a lower cardiovascular risk category and less likely to prescribe combination therapy in women compared to men of the same age with established cardiovascular disease. This reflects trends in the general population where women have worse outcomes following a cardiovascular event due to differences in their symptoms, presentations and pathophysiology. We found no evidence of significant sex differences in the cause of kidney failure by age (not presented). Systemic conditions such as SLE are captured as part of the glomerulonephritis/IgA nephropathy group.

Minor comments1. Some statements in the Introduction should include a supporting reference:a. “Growing evidence support sex-specific mechanisms….leading to systematic differences in care including misdiagnosis and delays in receiving treatment.”b. “Females are also under-represented in randomized clinical trials and animal models…”c. “Women may still be referred for specialist care later than men with comparable kidney function.”Response: We have now included further references to support these statements in the introduction.

2. Towards the end of the Introduction, the authors cite 2 previous papers (references 9 and 10) that have examined this question. They first state that the design of these studies “does not account for background mortality” but later cite the same 2 papers as examples of studies that “adjusted for background mortality.” Please clarify.Response: We have clarified that these studies mainly focus on the comparisons within the population with KF, providing little insight into differences in seen in the one relative measure reported in each study.

3. In Figures 3B and 3C, the p-values (<0.001) are different to those in the text (<0.01).Response: This was a typo, both are <0.001. We have amended the text.

4. The labels in Figure 3 refer to p-values for sex interactions. Are these age*sex and time*sex interactions?

Response: Figure 3A-3B are interaction between sex and age, and Figure 3C is interaction between sex and calendar year. We have added a footnote to Figure 3 to clarify.

Reviewer: 5Thank you for the opportunity to review this interesting and important work.I apologize for not having this done last week, when I had hoped to finish it. Thank you for your patience.This is a novel and comprehensive examination of absolute and relative differences mortality and other measures of survival, for patients on dialysis and those transplanted, for men and women. The dataset is well suited to the analysis and the paper is a significant contribution in the field.The paper is well written and easy to understand.I didn’t understand why the authors chose indirect standardization when I believe that they have the data needed for direct standardization.Response: We chose to use indirect standardization as it provides a measure of excess deaths relative to the general population, rather than direct standardization that produces a standardized rate that would still need reference to general population standardized rates to understand the excess burden.

SMRs were adjusted for 5-year age band, sex, calendar year and country; sensible choices.How was transplantation handled in the analysis of patients on dialysis, and return to dialysis for the analysis of patients with transplants? It’s a difficult issue because censoring would clearly be biased in both situations. There is also the additional issue that if access to transplantation were biased in some way, eg, against women, healthier women (who would have been transplanted had there been equity in decision making) will remain on dialysis contributing follow up in the dialysis group.Response: We have clarified in the methods that we took a once transplanted, always transplanted approach for our subpopulation of kidney transplantation recipients. We agree that it is difficult to disentangle the effects of returning to dialysis and mortality among this subpopulation. We believe this approach, analogously to ‘intention to ‘treat’, provides the best overview of a patient’s lived experienced after receiving a kidney transplant. We have also included in the discussion the possibility of access to transplantation being biased towards selecting healthier women, which aligns with our study findings of no excess mortality and comparable survival between women with kidney transplants and women in the general population.

Figure 1 – I would suggest including a second panel showing causes of death in the general population.Sex differences compared with the general population. As the authors discuss, mortality rates are similar between men and women across age (figure 3a). However, compared to the general population using SMRs (figure 3b) there is evidence of an interaction, most marked at younger ages, with SMRs in men 18-50 markedly different from those in women. The causes of death in these age groups in the general population ( https://www.aihw.gov.au/reports/life-expectancy-death/deaths-in-australia/contents/leading-causes-of-death) are very different from those reported here for patients on dialysis: top 3 are suicide, land transport accidents and accidental poisoning. I wondered if, given that these are the top causes, there might be large differences between rates in men and women in the general population that account for the interaction, rather than large differences between men and women on dialysis.Response: Causes of death is a minor part of this study, as the main focus is on the relative measures of survival given by the excess deaths, excess mortality ratio from relative survival and years of life lost (YLL). Approximately 50% of deaths are attributed to chronic kidney disease and diabetes, which is expected from the KF population but is vastly different to the causes of death in the general population. We believe presenting the causes of death from the general population would not aid the main findings from our paper and may confuse readers.

Hemo study. The reference for the causal-type statement ‘led to’ is the randomized trial (Eknoyan et al NEJM 2002) in which gender*dose was one of two prespecified tests for interaction that achieved conventional statistical significance: there were fourteen tests for interaction. Statistical significance was lost after Bonferroni correction. The paper cited is also relevant but is a cohort study. I think this discussion point could be written more precisely describing both these works; do the authors think we should put much weight on these findings?Response. We discussed the HEMO study once in the introduction of our paper. We have now also included the Eknoyan et al study to highlight these effects were lessened in a subsequent clinical trial. It is still interesting to note that the clinical trial found risk of death in females was 19% lower in the high-dose group, but the risk of death in males was 16% higher in the high-dose group (p=0.01). This effect was not diminished by confounding but instead due to the lower p-value from the Bonferroni method, which could suggest an issue of statistical power rather than lack of evidence for the use of higher dosages among women.

We have revised to discussion to not further mention these studies.

Discussion p12 l17-22. It is common to call sex disparities in transplantation ‘access’ to transplantation but I think it would be better not to. What is being reported is relative rates. Adjusted for PRA, RR transplant was 0.98 in the US study. There was a difference (0.84), as the authors comment, in waitlisting, that could not be accounted for by factors in the study, but since this was a study of administrative data, it is not able to distinguish the many different behaviours on the part of HCW and patients that led to the discrepancy. Again, more granularity in describing these studies would help the reader. The Canadian study, for example, is a study from 2000 of transplantation, the US study examined waitlisting and transplantation, as did the Australian study, which could be described in more detail because of the relevance to the population in the current study and the recency of the data.Response: We have amended this sentence to state “sex disparities in transplantation”. We have further elaborated on the findings of the Tong et al paper, which did found 2 of the 17 studies included evaluate the impact of sex. Women were more likely to be considered unsuitable for transplantation compared to men of the same age and comorbid conditions.

How censoring was handled (of failed transplants and for patients on dialysis undergoing transplantation) will warrant discussion. It’s hard to comment on this not knowing what was done.Response: “Death ascertainment and cause of death” section in the methods outlines that time at risk was measured from the date of kidney replacement therapy for the KF population and the earliest transplant date for the subpopulation of kidney transplant recipients, until the death of death or censored at 31st Dec 2019. Hence, this takes a once transplanted, always transplanted approach. We performed this approach analogously to an ‘intention to treat’ approach. Patients with KF often experience both dialysis modalities and kidney transplantation, and our analysis provides a lived experience of a typical patient’s journey for care of KF.

Similarly, I would suggest not using the interpretative word ‘equitable’ in line 29; survival was similar in both sexes without excess mortality in women; however since women in the general population live longer than men, this suggests that there is a sex-related difference (from expected patterns) in transplant patients also. Whether this is due to inequities is a strong hypothesis but is difficult to sort out in a study of this design.Response: We have removed the word equitable and replaced it with similar.

Reviewer: 6Apologies to the authors for the lateness of my review. An interesting study with interesting

findings. The methods generally look fine - I have some minors comments for clarifications on the Methods.

However, my overriding concern is the obvious one, relating to the age of the data. For the Australian data this goes from 1980 to the end of 2013 and for the New Zealand data, this goes from 1988 to the end of 2012. So clearly coverage for a good time span, but both are 7-8 years old now. Why wasn’t more contemporary data obtained? Unless the data are updated to include more recent data, I’m struggling with the usefulness of these findings for today – regardless how the interesting they are historically. Has there been any changes in the last 10 years or so that could change these findings/or make them less relevant?Response: We have addressed these concerns above. Briefly, when linking to national death registers there is a 2-3 year lag for death information. We have updated the fact of death until 2019 using ANZDATA, which has been shown to have a strong agreement with national death registers. There have been no major changes to our findings as a result of updated data. Our trends in excess deaths over time suggest the sex disparity has not continued to converge, but instead has stagnated in recent years where females continue to experience a greater burden of excess deaths. The excess mortality ratio (EMR) has also increased from 9% to 13%, where females have 13% excess mortality at any given time compared to males. We believe updated study data has strengthen our study and highlight the importance and robustness of our findings.

Page 9: 2nd paragraph/Figure 2. What analysis are these findings from? What is this line in Figure 2, a smoother sorts or a spline? If so, then this needs clarifying in the Methods or as a footnote to the Figure.Response: The percentages for causes of death are now compared using logistic regression to examine trends over time, as suggested by reviewer 2. We have added a footnote to Figure 2 to outline this is a lowess smoothed line.

Why were patients with missing smoking status excluded from the analysis? Why not impute – roughly 10% had missing smoking status.Response: We have clarified that smoking status was not routinely collected in ANZDATA until after 1991. As suggested, we have used multiple imputation chained equations to impute the smoking status with a multinomial logistic regression of 5 iterations. There were no major differences in the estimated life expectancy or years of life lost (YLL) as a result of the imputation.

How was age adjusted for? Table 1 has age summarized as categories, was this how it was adjusted for in the analysis? Or was is adjusted for on its original continuous scale? If so, was nonlinearity in age not considered, say using splines?Response: Table 1 summarizes the patient cohort. In the methods, we have clarified that the study population was matched to the general population based on 5-year age bands to estimate the standardized mortality ratios (SMR), relative survival, and years of life lost (YLL). We then produced SMRs by categorized age groups (as shown in Figure 3A-3B). For modelling the excess mortality ratio (EMR) we categorized age as given in Table 1. The parametric survival model to estimate life expectancy included time at risk relative to the patient’s date of birth, so the hazard of death was based on a patient’s age. The subsequent years of life lost was taken as the difference between the estimated life expectancy survival and the life expectancy from the general population, matching to 5-year age band at initiation of kidney replacement therapy, sex, calendar year and country. These are currently described in the methods and we have clarified that it is based on 5-year age bands when matching to the general population, categorized in the EMR model and as time at risk in the parametric survival model.

**Information for submitting a revision**

Deadline: Your revised manuscript should be returned within one month.

How to submit your revised article: Log into http://mc.manuscriptcentral.com/bmj and enter your Author Center, where you will find your manuscript title listed under "Manuscripts with Decisions." Under "Actions," click on "Create a Revision." Your manuscript number has been appended to denote a revision.

You will be unable to make your revisions on the originally submitted version of the manuscript. Instead, revise your manuscript using a word processing program and save it on your computer. Once the revised manuscript is prepared, you can upload it and submit it through your Author Center. When submitting your revised manuscript, you will be able to respond to the comments made by the reviewer(s) and Committee in the space provided. You can use this space to document any changes you make to the original manuscript and to explain your responses. In order to expedite the processing of the revised manuscript, please be as specific as possible in your response to the reviewer(s). As well as submitting your revised manuscript, we also require a copy of the manuscript with changes highlighted. Please upload this as a supplemental file with file designation ‘Revised Manuscript Marked copy’. Your original files are available to you when you upload your revised manuscript. Please delete any redundant files before completing the submission.

When you revise and return your manuscript, please take note of all the following points about revising your article. Even if an item, such as a competing interests statement, was present and correct in the original draft of your paper, please check that it has not slipped out during revision. Please include these items in the revised manuscript to comply with BMJ style (see: http://www.bmj.com/about-bmj/resources-authors/article-submission/article-requirements andhttp://www.bmj.com/about-bmj/resources-authors/forms-policies-and-checklists).

Items to include with your revision (see http://www.bmj.com/about-bmj/resources-authors/article-types/research:

1. What this paper adds/what is already known box (as described at http://resources.bmj.com/bmj/authors/types-of-article/research)

2. Name of the ethics committee or IRB, ID# of the approval, and a statement that participants gave informed consent before taking part. If ethics committee approval was not required, please state so clearly and explain the reasons why (see http://resources.bmj.com/bmj/authors/editorial-policies/guidelines.)

3. Patient confidentiality forms when appropriate (see http://resources.bmj.com/bmj/authors/editorial-policies/copy_of_patient-confidentiality).

4. Competing interests statement (see http://resources.bmj.com/bmj/authors/editorial-policies/competing-interests)

5. Contributorship statement+ guarantor (see http://resources.bmj.com/bmj/authors/article-submission/authorship-contributorship)

6. Transparency statement: (see http://www.bmj.com/about-bmj/resources-authors/forms-policies-and-checklists/transparency-policy)

7. Copyright statement/licence for publication (see http://www.bmj.com/about-bmj/resources-

authors/forms-policies-and-checklists/copyright-open-access-and-permission-reuse)

8. Data sharing statement (see http://www.bmj.com/about-bmj/resources-authors/article-types/research)

9. Funding statement and statement of the independence of researchers from funders (see http://resources.bmj.com/bmj/authors/article-submission/article-requirements).

10. Patient and public involvement statement

https://docs.google.com/document/d/1djgVLEUFtPQzLpf5HyuiFcMgrNJ_o7Yb8z73zgXxXYM/e

11. Dissemination plans: At the end of the paper please state how the results of your study have been (or will be) sent to patients and the public under the heading “Dissemination plans”. If you have prepared a lay summary eg for your funders, please include it in a supplementary file.

If you have not disseminated and have no plans to do so, please state why.

12. Patient confidentiality forms when appropriate

13. Please ensure the paper complies with The BMJ’s style, as detailed below:

a. Title: this should include the study design eg "systematic review and meta-analysis.”

b. Abstract: Please include a structured abstract with key summary statistics, as explained below (also see http://resources.bmj.com/bmj/authors/types-of-article/research). For every clinical trial - and for any other registered study- the last line of the abstract must list the study registration number and the name of the register.

Please report all outcomes that were listed in the trial registry, or explain that you will publish them elsewhere. Please clearly identify each outcome as primary, secondary, or post-hoc in the text, abstract, and any tables or figures. We expect authors to report prespecified outcomes. If outcomes in the trial registry have later been changed, please explain the reasons for the change and the dates of the change in the paper. You may report the changed outcomes, but we will expect you to also report on the originally specified outcomes unless otherwise agreed with the handling editor for your paper.

Occasionally the outcomes that are prespecified in a trial registry do not match up with those included in the trial protocol. When there are discrepancies between protocol and registry specified outcomes, we expect the paper to report and interpret the registry specified outcomes. You may also report any protocol specified outcomes, but if you do please be sure to include the date of the protocol and the point at which each outcome was added to the protocol, and explain why the registry entry differed from the protocol and why the registry was not updated to reflect any protocol changes.

c. Introduction: This should cover no more than three paragraphs, focusing on the research question and your reasons for asking it now.

d. Methods: For an intervention study the manuscript should include enough information about the intervention(s) and comparator(s) (even if this was usual care) for reviewers and readers to understand fully what happened in the study. To enable readers to replicate your work or implement the interventions in their own practice please also provide (uploaded as one or more supplemental files, including video and audio files where appropriate) any relevant detailed descriptions and materials. Alternatively, please provide in the manuscript urls to openly accessible websites where these materials can be found.

e. Results: Please report statistical aspects of the study in line with the Statistical Analyses and Methods in the Published Literature (SAMPL) guidelines http://www.equator-network.org/reporting-guidelines/sampl/. Please include in the results section of your structured abstract (and, of course, in the article's results section) the following terms, as appropriate:

i. For a clinical trial: Absolute event rates among experimental and control groups; RRR (relative risk reduction); NNT or NNH (number needed to treat or harm) and its 95% confidence interval (or, if the trial is of a public health intervention, number helped per 1000 or 100,000.)ii. For a cohort study: Absolute event rates over time (eg 10 years) among exposed and non-exposed groups; RRR (relative risk reduction.)iii. For a case control study:OR (odds ratio) for strength of association between exposure and outcome.iv. For a study of a diagnostic test: Sensitivity and specificity; PPV and NPV (positive and negative predictive values.)v. For a systematic review and/or meta-analysis: Point estimates and confidence intervals for the main results; one or more references for the statistical package(s) used to analyse the data, eg RevMan for a systematic review. There is no need to provide a formal reference for a very widely used package that will be very familiar to general readers eg STATA, but please say in the text which version you used. For articles that include explicit statements of the quality of evidence and strength of recommendations, we prefer reporting using the GRADE system.

Please report all outcomes that were listed in the trial registry, or explain that you will publish them elsewhere. Please clearly identify each outcome as primary, secondary, or post-hoc in the text, abstract, and any tables or figures. We expect authors to report prespecified outcomes. If outcomes in the trial registry have later been changed, please explain the reasons for the change and the dates of the change in the paper. You may report the changed outcomes, but we will expect you to also report on the originally specified outcomes unless otherwise agreed with the handling editor for your paper.

Occasionally the outcomes that are prespecified in a trial registry do not match up with those included in the trial protocol. When there are discrepancies between protocol and registry specified outcomes, we expect the paper to report and interpret the registry specified outcomes. You may also report any protocol specified outcomes, but if you do please be sure to include the date of the protocol and the point at which each outcome was added to the protocol, and explain why the registry entry differed from the protocol and why the registry was not updated to reflect any protocol changes.

f. Discussion: To minimise the risk of careful explanation giving way to polemic, please write the discussion section of your paper in a structured way. Please follow this structure: i) statement of principal findings of the study; ii) strengths and weaknesses of the study; iii) strengths and weaknesses in relation to other studies, discussing important differences in results; iv) what your study adds (whenever possible please discuss your study in the light of relevant systematic reviews

and meta-analyses); v) meaning of the study, including possible explanations and implications for clinicians and policymakers and other researchers; vi) how your study could promote better decisions; vi) unanswered questions and future research

g. Footnotes and statements

Online and print publication: All original research in The BMJ is published with open access. Our open access policy is detailed here: http://www.bmj.com/about-bmj/resources-authors/forms-policies-and-checklists/copyright-open-access-and-permission-reuse. The full text online version of your article, if accepted after revision, will be the indexed citable version (full details are at http://resources.bmj.com/bmj/about-bmj/the-bmjs-publishing-model).