Estimated glomerular filtration rate predicts 30-day mortality in medical emergency departments: Results of a prospective multi-national observational study.

BACKGROUND: Renal failure is common in patients seeking help in medical emergency departments. Decreased renal function is associated with increased mortality in patients with heart failure or sepsis. In this study, the association between renal function (reflected by estimated glomerular filtration rate (eGFR) at the time of admission) and clinical outcome was evaluated. METHODS/
OBJECTIVES: Data was used from a prospective, multi-national, observational cohort of patients treated in three medical emergency departments of tertiary care centers. The eGFR was calculated from the creatinine at the time of admission (using the Chronic Kidney Disease-Epidemiology Collaboration equation,CKD-EPI). Uni- and multivariate regression models were used for eGFR and 30-day mortality, in hospital mortality, length of stay and intensive care unit admission rate.
RESULTS: 6983 patients were included. The 30-day mortality was 1.8%, 3.5%, 6.9%, 11.1%, 13.6%, and 14.2% in patients with eGFR of above 90, 60-89, 45-59, 30-44, 15-29, and <15 ml/min/1.73m2, respectively. Using multivariate regression, the adjusted odds ratio (OR) was 2.31 (for 15-29 ml/min/1.73m2, 95% confidence interval 1.36 to 3.90, p = 0.002) and 3.73 (for eGFR <15ml/min/1.73m2 as compared to >90 ml/min/1.73m2, 95% CI 2.04 to 6.84, p<0.001). For 10 ml/min/1.73m2 decrease in eGFR the OR for the 30-day mortality was 1.15 (95% CI1.09 to 1.22, p<0.001).The eGFR was also significantly associated with in-hospital mortality, the percentage of ICU-admissions, and with a longer hospital stay. No association was found with hospital readmission within 30 days. As limitations, only eGFR at admission was available and the number of patients on hemodialysis was unknown.
CONCLUSION: Reduced eGFR at the time of admission is a strong and independent predictor for adverse outcome in this large population of patients admitted to medical emergency departments.

Introduction

Impaired renal function represents a diagnostic and therapeutic challenge for the emergency department [1]. Independent of the cause of admission, renal insufficiency might impact the disease courses in several ways. It potentially adds disturbances to volume status, water, acid base, and electrolytes as clinical challenges. Renal failure leads to a proinflammatory but also immunocompromised state, which promotes injury to distal organs including lung, heart and nervous system [2-5]. Kidney function might influence the interpretation of diagnostic tests and biomarkers (e.g troponins and natriuretic peptides) [6]. Finally, dosing of various drugs has to be adjusted, which entails the danger of both overdosing resulting in toxicity, and underdosing with insufficient therapeutic effect [7, 8]

Decreased renal function in both acute kidney injury, as well as chronic kidney disease has been shown to be associated with increased in-hospital mortality, increased resource utilization, and increased hospital cost of care [9-15]. For example the outcome in patients with sepsis or heart failure is significantly worse in those with kidney disease [9, 13, 16–20].

We hypothesize that impaired renal function, reflected by eGFR (CKD-EPI) from admission creatinine at time of medical emergency department (ED) presentation, might be a marker of poor outcome. We used data from a large, multi-national, prospective, observational study, initially designed to investigate triage prioritization improvement by the addition of biomarkers [21], to evaluate associations with clinical outcome.

Methods

Study design

We used data of the TRIAGE study. The purpose of this study was to evaluatewhether the addition of biomarkers would improve early risk stratification in a “real life” population of medical patients presenting to the ED of tertiarycare hospitals. Hier meinen Comment zu Rev2 Frage 3 einfügen?. A total of 7132 consecutive patients were recruited from March 2013 to October 2014 [21, 22]. After excluding patients with missing serum creatinine values, a cohort of 6,983 patients was included in the current investigation (4,554 from Aarau (Switzerland), 1,460 from Paris (France), and 969 from and Clearwater (FL, USA). The baseline characteristics per center are included in the supplemental data ().

As an observational, quality control study, the Institutional Review Boards of the three hospitals approved the study and waived the need for individual informed consent (Ethikkommission Kanton Aargau (EK 2012/059); CCTIRS-Comité consultatif sur le traitement de l’information en matière de recherche; CPP ID RCB:2013-A00129-36) MPM-SAH Institutional ReviewBoard, Clearwater; FL,IRB number 2013_005). The study was registered at “ClinicalTrials.gov” registration website and the study protocol has been published [21].

Data collection and definitions of diagnoses

Laboratory evaluation was only included from the time of ED admission and was part of the routine workup. All participants provided a medical history and underwent a physical examination. Congestive heart failure, coronary heart disease, hypertension, chronic obstructive pulmonary disease, dementia, diabetes mellitus, cancer, renal failure and history of stroke were documented as co-morbidities. Based on the information at discharge from the ED, patients were grouped into main diagnosis groups (e.g. infectious, cardiovascular) by two independent physicians. Management throughout the hospital stay was at the discretion of the treating physician, independent of the research team. A standardized telephone interview was performed 30 days after hospital admission to assess vital status, functional outcome and hospital readmissions.

Primary and secondary endpoints

The primary endpoint was all-cause 30-day mortality. Secondary endpoints were in-hospital mortality, admission to the intensive care unit (ICU), hospital readmission within 30 days, and length of hospital stay (LOS). The decision for ICU admission was left to the discretion of the treating physicians.

Assessment of renal function

Serum creatinine was measured by the clinical chemistry laboratory photometrically using the Jaffe reaction (Siemens, Dimension Vista® System, Flex ® reagent cartridge CREA). The eGFR was calculated according to the CKD-EPI 2009 equation expressed for specified sex and serum creatinine level as follows [23]:

GFR = 141 × min (Scr/κ, 1)α × max(Scr/κ, 1)-1.209 × 0.993Age × 1.018 [if female].

Scr is serum creatinine in μmol/L, κ is 61.9 for females and 79.6 for males, α is -0.329 for females and -0.411 for males, min indicates the minimum of Scr/κ or 1, and max indicates the maximum of Scr/κ or 1. No adjustment was performed for black or African ethnicity as the information was not available from the data collection.

The eGFR was categorized into six categories (eGFR >90 ml/min/1.73 m2, 60–89 ml/min/1.73 m2, 45–59 ml/min/1.73 m2, 30–44 ml/min/1.73 m2,15–29 ml/min/1.73 m2, <15 ml/min/1.73 m2) as proposed by the Kidney Disease Improving Global Outcomes (KIDIGO) Guidelines 2012 for chronic kidney diseases [24].

Statistical analysis

Descriptive statistics, including mean with standard deviation (SD), median with interquartile range (IQR), and frequencies were used to describe the study population, when appropriate. To evaluate associations of eGFR with primary and secondary endpoints univariate and multivariate logistic regression analyses were applied and odds ratios (OR) with 95% confidence intervals (CI) were reported. The models were adjusted in a stepwise algorithm for age, gender, main diagnosis and comorbidities. The discriminatory power was described using area under the receiver operating characteristics curve analysis (AUC). All tests were two-tailed and carried out at 5% significance levels (exceptions stated separately). Analyses were performed with Stata 12.1 (Stata Corp., College Station, TX, USA).

Results

Patient characteristics

A total of 6,983 patients with follow-up data were included in the analysis. The median age was 62 years and 46.7% of patients were female (). The three most frequent diagnoses were cardiovascular diseases (23.4%), neurological diseases (22.2%) and infections (14.7%). Patients had a high burden of comorbidities including hypertension (39.7%), coronary heart disease (11.9%), diabetes (15.4%), and cancer (13.7%). Most patients were treated as inpatients (72.8%).

Associations of eGFR with primary and secondary endpoint

A total of 325 Patients (4.7%) died within thirty days of admission. The percentage was 1.8% in patients with an eGFR >90 ml/min/1.73m2 compared to 14.2% in patients with an eGFR <15ml/min/1.73m2 ().

Primary and secondary outcomes stratified by eGFR groups and with eGFR as a continuous variable are illustrated in Tables . In the unadjusted model for 30-day mortality the OR was 1.94 (95% CI 1.35 to 2.77, p<0.001) in the eGFR group of 60–89 ml/min/1.73 m2 and with 8.80 (95% CI 5.27 to 14.68, p<0.001) significantly higher for the eGFR group <15ml/min/1.73m2.

After adjustment for important confounders (such as age, gender, main diagnosis, and comorbidities) the groups with an eGFR below 45 ml/min/1.73m2 remained significantly associated with 30-day mortality (). Using multivariate regression the adjusted OR was 1.70 (95% CI, 1.05 to 2.75, p = 0.03) for 30–44 ml/min/1.73m2, 2.31 (95% CI 1.36 to 3.90, p = 0.002) for 15-29ml/min/1.73m2 and 3.73 (95% CI 2.04 to 6.84, p<0.001) for the lowest group (<15ml/min/1.73m2), as compared to the highest eGFR group.

The in-hospital mortality with eGFR <15 ml/min/1.73m2 was lower than in the group with a eGFR of 15–30 ml/min/1.73m2, with a relative small number of patients and events in both groups.

Each decrease in eGFR of 10 ml/min/1.73m2 was associated with a 15% increased risk of 30-day mortality (OR 1.15, 95% CI 1.09 to 1.22, p<0.001) ().When adjusted for the selected biomarkers Hemoglobin (Hb), C-reactive Protein (CRP), and Copeptin individually, eGFR remained significantly associated with 30-day mortality. When adjusted for all biomarkers at once the association with eGFR was no longer apparent ().

There was a ‘U’-shaped association between eGFR and 30-day mortality, with a nadir at approximately 130 ml/min/1.73m2 (). The curve was relatively flat between 100 and 160 ml/min/1.73 m2, but increased sharply at lower and higher levels of eGFR (with the higher levels being outside the physiologic meaningful area. These are not illustrated in the figure). Kaplan-Meier curves show a stepwise increase in mortality in the different eGFR classes ().

Hazard ratio of 30-day mortality as a function of eGFR.

(x-axis, eGFR (ml/min/1.73m2), y-axis, odd ratio for 30-day mortality, CI, confidence interval).

As secondary endpoints the in-hospital mortality, ICU admission, readmission and length of hospital stay were further evaluated.

A decrease in eGFR was significantly associated with in-hospital mortality, ICU-admission, and with a prolonged length of stay (). After adjustment for confounders we found a significantly higher in-hospital mortality in patients with an eGFR <30 ml/min/1.73 m2, as compared to the reference group.

In contrast, even mild impairment in renal function (<90 ml/min/1.73 m2) was significantly associated with an increased risk for ICU admission and longer hospital stay ().

For every drop of 10 mL/min/1.73 m2 the adjusted OR was 1.15 (95% CI, 1.10 to 1.2, p<0.001), for ICU admission and 1.16 (95% CI1.08 to 1.24, p<0.001) for in-hospital mortality. LOS increased by 0.25 days for every drop of 10 mL/min/1.73 m2 (95% CI 0.19 to 0.32, p<0.001) in the adjusted model ().

There was no significant association between a decrease in eGFR and hospital readmission rate.

Discriminative performance of eGFR

The receiver operating characteristic (ROC) curve showed acceptable discrimination by eGFR with an AUC of 0.71 (95% CI 0.68 to 0.73) for 30-day mortality and an AUC of 0.71 (95% CI 0.68 to 0.75) for in-hospital mortality in the total cohort ().

Discussion

This study illustrates that eGFR, (using the CKD-EPI equation) from a single serum-creatine measurement, was significantly associated with adverse clinical outcomes, particularly reflected by the 30-day mortality, as well as with in-hospital mortality, ICU admission, and LOS. Decreased eGFR might therefore serve as a general risk marker as reflected in this largest cohort of medical emergency patients studied so far.

The data is consistent with previous studies which described an association between poor outcome and decreased renal function in particular disease entities. These include patients with myocardial infarction [25, 26], heart failure [9, 10, 27], those undergoing percutaneous cardiovascular interventions [28], coronary artery bypass graft surgery [11, 29], and stroke [30, 31]. Our approach was to evaluate the eGFR as a general risk marker irrespective of the final diagnosis for daily routine in emergency care.

It is currently not clear whether poor renal function at time of presentation is just a marker for sicker patients or whether there is a causal relationship with the increased mortality. Several factors might be involved in the higher mortality such as electrolyte disturbances, metabolic acidosis, volume overload, anemia and the negative impact of uremic compounds (resulting in e.g. vasculopathy or increased risk for infection).

We would suggest that in patients with mildly reduced renal function the eGFR reduction might be a reflection of a sicker patient population, whereas in severe renal failure the complexity of the treatment will increase significantly, and the causality hypothesis might be true.

The data illustrates a significantly higher mortality risk for the eGFR 30–44 ml/min/1.73m2 group as compared to the group with an eGFR 45–60 ml/min/1.73m2.This supports the decision of the KDIGO to subdivide stage 3 in chronic kidney disease into categories a and b (as also reflected by the risk in the general population) [24].

Patients with an eGFR 15–29 ml/min/1.73m2 had a worse in-hospital mortality as compared with patients with a eGFR below 15 ml/min/1.73m2 (Table 3, Fig 2). This is a surprising finding without a clear explanation. The relatively small patient groups might have led to this result. Both the 30-day mortality and ICU admission rate are higher in patients with an eGFR 15 ml/min/1.73m2 which indicates that this group is actually sicker than the afore mentioned. In-hospital mortality is influenced by hospital transfer policies and therefore the outcome at defined time points (e.g. 30, 60, 90 days) might be the more robust parameter [32, 33]. Our data illustrates a ‘U’-shaped relationship between eGFR and all-cause mortality, with increased risk among those with both low and very high values of eGFR, respectivly. This is in line with the findings of Shlipak et al. [34]. They showed that the association between quintiles of creatinine and all-cause mortality appeared to be J-shaped among 4,637 participants in the Cardiovascular Health Study. An elevated risk of cardiovascular events was also found among patients with atherosclerotic cardiovascular disease with an eGFR >125 ml/min/1.73 m2 by Inrig et al. [35]. The eGFR might be elevated due to hyperfiltration in early diabetic nephropathy, diet (e.g. dietary supplements, vegetarians), medication use and rapidly changing kidney function [30]. Patients with abnormal muscle mass might also be included (e.g. amputation, muscular disease, chronically ill).

Table 3

Univariate and multivariate logistic regression analyses according to continuous eGFR values.

	Unadjusted model	Model 1	Model 2	Model 3
	OR (95% CI), p-value
30-day mortality
eGFR per decrease 10ml/min/1.73m2	1.28 (1.24 to 1.33), p<0.001	1.15 (1.09 to 1.20), p<0.001	1.14 (1.09 to 1.20), p<0.001	1.15 (1.09 to 1.22), p<0.001
In-hospital mortality
eGFR per decrease 10ml/min/1.73m2	1.29 (1.23 to 1.36), p<0.001	1.16 (1.09 to 1.24), p<0.001	1.16 (1.08 to 1.23), p<0.001	1.16 (1.08 to 1.24), p<0.001
ICU admission
eGFR per decrease 10ml/min/1.73m2	1.16 (1.13 to 1.20), p<0.001	1.16 (1.12 to 1.21), p<0.001	1.16 (1.12 to 1.21), p<0.001	1.15 (1.10 to 1.20), p<0.001
Rehospitalisation
eGFR per decrease 10ml/min/1.73m2	1.06 (1.03 to 1.09), p<0.001	1.05 (1.01 to 1.09), p = 0.015	1.05 (1.01 to 1.09), p = 0.018	1.03 (0.99 to 1.07), p = 0.132
LOS *
eGFR per decrease 10ml/min/1.73m2	0.5 (0.45 to 0.55), p<0.001	0.30 (0.24 to 0.37), p<0.001	0.30 (0.23 to 0.36), p<0.001	0.25 (0.19 to 0.32), p<0.001

95% CI 95% Confidence interval, eGFR estimated glomerular filtration rate, ICU intensive care unit, LOS length of stay, OR Odds ratio, Models were stepwise adjusted for age (model 1), age, and gender (model 2), age, gender, main diagnosis, and comorbidities (model 3)

*Since LOS is a continuous variable regression results represent a regression coefficient (95% confidence interval)

Fig 2

Kaplan-Meier survival estimates by eGFR groups.

x-axis, day since admission, y-axis, Proportion of patients alive, eGFR, estimated glomerular filtration rate (ml/min/1.73m2).

To the best of our knowledge, this is the largest study using eGFR calculated with the CKD-EPI formula in patients presenting to the medical ED. There has been intensive discussion about which formula might be most suitable in specific situations. For example, Moreno et al. found that the Cockcroft-Gault formula was a slightly better predictor of mortality in acute heart failure patients [20]. The Cockcroft-Gault formula contains the body weight of the patient, a variable that might provide additional information about patients’ constitution or nutritional state and might have influenced the shape of our eGFR vs. mortality curve. Unfortunately, as data on body weight were not available in our cohort, we were not able to compare these two.

Several limitations of our study need to be discussed. First, this is a secondary analysis of a previous observational study. We are therefore limited to the parameters collected in the primary study. Important factors which are not available are the cause of death, the course of renal function (i.e. acute kidney injury versus chronic kidney disease), and the number of patients on dialysis.

The single measurement of serum creatinine at hospital admission cannot discriminate between acute kidney injury, acute kidney disease and chronic kidney disease. Moreover, eGFR levels might change over the course of a patients stay. It would have been an interesting question whether the risk differs in acute versus chronic renal impairment, and also whether improvement of renal function during the hospital stay would be associated with better outcome.

As dialysis patients might bias the group with an eGFR <15 ml/min/1.73m2, an analysis was performed without these patients potentially on hemodialysis. The eGFR remained significantly associated with 30-day mortality, both as a continuous variable, and for patients with an eGFR 30–45 ml/min/1.73m2 and 15–30 ml/min/1.73m2 (). Although, the number of dialysis patients in this analysis might be higher than in the general population, the absolute number is expected to be low (but still an important source of bias). Whether a risk stratification with eGFR would lead to better patient care and to a lowering of adverse outcome in this unselected population cannot be concluded. Causality cannot be implied by our data. We hope to create further hypotheses and result in studies with detailed analysis of the course of renal failure.

Conclusion

Decreased renal function reflected by a reduction in eGFR (CKD-EPI), at time of ED admission was associated with increased risk of 30-day mortality, ICU admission, and length hospital stay. The eGFR may serve as a useful tool for risk stratification in the medical emergency department.

Baseline characteristics of the total cohort by centers.

(DOCX)

Click here for additional data file.

Associations of eGFR with adverse clinical outcome in an univariate model and after adjustment for laboratory results.

(DOCX)

Click here for additional data file.

Associations of eGFR with adverse clinical outcome in univariate and multivariate models without patients with a eGFR<15 ml/min/1.73m2.

(DOCX)

Click here for additional data file.

26 Sep 2019

PONE-D-19-23285

Estimated glomerular filtration rate predicts 30-day mortality in medical emergency departments: results of a prospective multi-national observational study

PLOS ONE

Dear Mr Haas,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

We would appreciate receiving your revised manuscript by Nov 10 2019 11:59PM. When you are ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter.

To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). This letter should be uploaded as separate file and labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. This file should be uploaded as separate file and labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. This file should be uploaded as separate file and labeled 'Manuscript'.

Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

We look forward to receiving your revised manuscript.

Kind regards,

Tatsuo Shimosawa, M.D., Ph.D.

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

http://www.journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and http://www.journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

1. Thank you for including your competing interests statement; "none"

Please complete your Competing Interests on the online submission form to state any Competing Interests. If you have no competing interests, please state "The authors have declared that no competing interests exist.", as detailed online in our guide for authors at http://journals.plos.org/plosone/s/submit-now

This information should be included in your cover letter; we will change the online submission form on your behalf.

Please know it is PLOS ONE policy for corresponding authors to declare, on behalf of all authors, all potential competing interests for the purposes of transparency. PLOS defines a competing interest as anything that interferes with, or could reasonably be perceived as interfering with, the full and objective presentation, peer review, editorial decision-making, or publication of research or non-research articles submitted to one of the journals. Competing interests can be financial or non-financial, professional, or personal. Competing interests can arise in relationship to an organization or another person. Please follow this link to our website for more details on competing interests: http://journals.plos.org/plosone/s/competing-interests

2. We note that you have stated that you will provide repository information for your data at acceptance. Should your manuscript be accepted for publication, we will hold it until you provide the relevant accession numbers or DOIs necessary to access your data. If you wish to make changes to your Data Availability statement, please describe these changes in your cover letter and we will update your Data Availability statement to reflect the information you provide.

3. Please amend your authorship list in your manuscript file to include author Beat Müller

4. Please amend the manuscript submission data (via Edit Submission) to include author Beat Mueller

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Partly

Reviewer #2: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: I Don't Know

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The authors studied the outcome of patients in the ED according to eGFR by the CKD-EPI formula.

1) Primary and secondary outcomes were adjusted for age, gender, main diagnosis, and comorbidities. Why were laboratory data not adjusted for? It seems that those with lower eGFR had higher CRP, copeptin and lower Hb which are probably related to the outcomes.

2) The proportion and basic characteristics of patients recruited from the three tertiary care centers should be mentioned (perhaps as supplementary data).

3) Limitation of not having data on race should be mentioned since this will incorrectly calculate eGFR and race is also probably an important confounding factor of the outcome. Is data on race really not available at least for the cohort from the US?

4) I am not really convinced that this study states a novel insight because serum Cr or eGFR has already been shown to be a predictive factor in chronic and acute settings. Mixing the chronic and acute setting will probably yield thee same results. Can you be more specific on how the result or method of this study is new?

Reviewer #2: Haas and colleagues evaluated the predictive performance of eGFR measured at ED for 30 day mortality. They found a step wise increase of mortality with reduction of eGFR, so eGFR seems to be a good predictor.

1) Were end-stage renal disease patients treated by maintenance dialysis included? If so, their eGFR should be different between before and after dialysis. This will cause a serious bias.

2) Cat 4 (eGFR 15-29) showed worse survival curve just before day 30 compared with Cat 5 (eGFR<15) (figure 2). The reason for this should be investigated. Did Cat 5 include many dialysis patients??

3) It is a little bit surprising that most (72.8%) pts were treated as inpatients in this ED cohort. Can the severity of the enrolled patients be evaluated? Can qSOFA score be caluculated?

4) As mentioned in the limitation section, this study did not distinguish AKI from CKD. However, because 72% of the enrolled patients admitted, subanalysis regarding AKI, CKD, acute-on-chronic can be performed when limited to the inpatients. This analysis will provide more precise role of single measurement of eGFR at ED.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Hiroo Kawarazaki

Reviewer #2: Yes: Kento Doi

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files to be viewed.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org. Please note that Supporting Information files do not need this step.

6 Nov 2019

Mr. Tatsuo Shimosawa, M.D., Ph.D.

Academic Editor, PLOS ONE

Basel, the 5th of November, 2019

Dear Doctor Shimosawa,

Thank you, for the opportunity to submit a revised manuscript of our study “Estimated glomerular filtration rate predicts 30-day mortality in medical emergency departments: results of a prospective multi-national observational study.” ID: PONE-D-19-23285) for publication in PLOS ONE. We appreciate the time invested in the manuscript by the editorial team and particularly the referees. The suggestions helped us to significantly improve the manuscript. The response to the question of the editorial team and the referees are listed in the point to point answer below.

We hope that the manuscript has reached publication quality for PLOS ONE.

Best regards

Laurent Haas, for all the co-authors

To Reviewer #1:

Dear Doctor Kawarazaki,

Thank you very much for the significant time, which you put into the evaluation of our manuscript. We appreciate your constructive comments, which helped us to improve the manuscript. We have incorporated your comments as follows.

Best regards

Laurent Haas.

1) Primary and secondary outcomes were adjusted for age, gender, main diagnosis, and comorbidities. Why were laboratory data not adjusted for? It seems that those with lower eGFR had higher CRP, copeptin and lower Hb which are probably related to the outcomes.

According to your suggestion we performed another analysis adjusting for the laboratory markers and added a table (Table A2) in the supplementary material.

When adjusted for selected variables (Hb, CRP, copeptin), the statistical significance of our main explanatory variable of interest, eGFR, becomes less strong. This suggests that copeptin and eGFR capture some of the same variation with respect to our outcome variable.

2) The proportion and basic characteristics of patients recruited from the three tertiary care centers should be mentioned (perhaps as supplementary data).

We have added the number of patients per country/center in the results section of the main manuscript (revised manuscript Page 8 paragraph 1).

Moreover, we have added patient characteristic stratified by center as supplementary data (Supplementary Table A4).

3) Limitation of not having data on race should be mentioned since this will incorrectly calculate eGFR and race is also probably an important confounding factor of the outcome. Is data on race really not available at least for the cohort from the US?

Race has of course a significant influence on eGFR, but this information was unfortunately not entered in the utilized database. We regret this limitation with regard to our data. We have added a note of caution and further discussed that in section of the limitation of the study (revised manuscript page 18 paragraph 2).

4) I am not really convinced that this study states a novel insight because serum Cr or eGFR has already been shown to be a predictive factor in chronic and acute settings. Mixing the chronic and acute setting will probably yield thee same results. Can you be more specific on how the result or method of this study is new?

Thank you for this important comment. As outlined in the introduction of our manuscript multiple studies assessing the predictive power of SCr and eGFR in a chronic setting. We are not aware of any larger studies of emergency patients. Given the necessity for a fast and easy (cost-efficient) biomarker for risk stratification in an emergency setting, we believe that it is worthwhile to assess its aptitude with respect to this setting. But it is correct that for the diseases already studied, our data is confirmative in nature. This was discussed in more detail (page 4 paragraph 3).

Reviewer #2:

Dear Doctor Kent Doi,

We appreciate your time and effort which you put into the evaluation of our manuscript. We think that these helped us to significantly improve the revised manuscripted. We have incorporated your comments as follows.

Best regards

Laurent Haas.

1) Were end-stage renal disease patients treated by maintenance dialysis included? If so, their eGFR should be different between before and after dialysis. This will cause a serious bias.

As rightly suggested chronic hemodialysis HD patients have highly different creatinine values before and after dialysis, and therefore a high variance in eGFR. Hemodialysis patients were included in our study although this variable was not surveyed, so our models could not be adjusted for this circumstance. The exact proportion of this patients in the cohort is unknown. We assume that although it is higher than in the general population, the absolute quantity should be relatively low. Dialysis patients have an increased mortality rate. This should also be reflected in the data with high baseline creatinine (even after dialysis) compared to the general population. To address your comment we have added a table (Table A3) with our calculations excluding patients with Cat 5 patients (as this group contains patients on dialysis) in the supplementary materials. These results are described in in the revised manuscript on page 16 paragraph 4). The results remain robust after excluding Cat 5 patients.

2) Cat 4 (eGFR 15-29) showed worse survival curve just before day 30 compared with Cat 5 (eGFR<15) (Figure 2). The reason for this should be investigated. Did Cat 5 include many dialysis patients?

This is indeed a curious finding, as the Cat. 4 patients seem to have a higher in hospital mortality but a lower 30-day mortality. We added the following paragraph in our manuscript (Page 16, Paragraph 3) Cat.5. patients have a higher ICU admission rate. These patients could be perceived as “sicker” and more short-term medical resources might be attributed (as reflected by ICU admission) to them, influencing discharge and transfer practices. Experts are discussing in-hospital vs 30-day mortality for outcome prediction/quality assessment and 30-day mortality may be a more valid measure, since it is a fixed point in time. In-hospital mortality is primary influenced by hospital transfer policies and long term (30, 60, 90 days) is mainly determined by diagnosis [1, 2]. Mortality seems to “catch up” with the Cat.5 patients on day 30.

The number of hemodialysis patients has been addressed under the first question.

3) It is a little bit surprising that most (72.8%) pts were treated as inpatients in this ED cohort. Can the severity of the enrolled patients be evaluated? Can qSOFA score be calculated?

We now provide frequency of inpatient care in the newly added Table A2 in the supplementary data (please see also our answer to question 2 of Reviewer #1). As you can see, there was a broad difference in frequency of inpatient care in the participating hospitals. This can be explained by the different health care systems studied. In the US-hospital and in the Swiss hospital medical patients are admitted through the emergency department for inpatient care, and patients with high probability of outpatient care are seen in a separate department (outpatient emergency service). In the French hospital, all of the patients are seen in the same department, which explains the distinct difference in inpatient treatment.

Unfortunately, data for respiratory rate is limited. That’s why we are not able to calculate SOFA score in a meaningful number of patients.

4) As mentioned in the limitation section, this study did not distinguish AKI from CKD. However, because 72% of the enrolled patients admitted, subanalysis regarding AKI, CKD, acute-on-chronic can be performed when limited to the inpatients. This analysis will provide more precise role of single measurement of eGFR at ED.

Indeed, most patients were treated as inpatients. In the database, there is only a one-time creatinine measurement available to us. During hospitalization creatinine monitoring was certainly performed but it was not integrated in the database, as the initial study design was to analyze the predictive power of novel biomarkers during triage in the emergency department. A follow up creatinine would have been interesting to differentiate between AKI and CKD. This would have provided additional information to the data and value to the paper.

We are aware of the limitations of the data; however, we believe that our data provide interesting information due to its large sample size and real-life cohort. It is a good reflection of the diverse patient population encountered in daily practice. We have emphasized this limitation in the revised manuscript (page 17 paragraph 3).

References:

1. van Rijn M, Buurman BM, Macneil Vroomen JL, Suijker JJ, ter Riet G, Moll van Charante EP, et al. Changes in the in-hospital mortality and 30-day post-discharge mortality in acutely admitted older patients: retrospective observational study. Age and Ageing. 2016;45(1):41-7. doi: 10.1093/ageing/afv165.

2. Vasilevskis EE, Kuzniewicz MW, Dean ML, Clay T, Vittinghoff E, Rennie DJ, et al. Relationship between discharge practices and intensive care unit in-hospital mortality performance: evidence of a discharge bias. Med Care. 2009;47(7):803-12. Epub 2009/06/19. doi: 10.1097/MLR.0b013e3181a39454. PubMed PMID: 19536006.

Submitted filename: Response to Reviewers.docx

Click here for additional data file.

29 Nov 2019

PONE-D-19-23285R1

Estimated glomerular filtration rate predicts 30-day mortality in medical emergency departments: results of a prospective multi-national observational study

PLOS ONE

Dear Mr Haas,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

The Policy of PLoS One request authors to provide all the data available.

We would appreciate receiving your revised manuscript by Jan 13 2020 11:59PM. When you are ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter.

To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). This letter should be uploaded as separate file and labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. This file should be uploaded as separate file and labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. This file should be uploaded as separate file and labeled 'Manuscript'.

Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

We look forward to receiving your revised manuscript.

Kind regards,

Tatsuo Shimosawa, M.D., Ph.D.

Academic Editor

PLOS ONE

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: (No Response)

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: No

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Just one minor comment. Table A4 has systolic BP and diastolic BP the other way round. All comments have been addressed.

Reviewer #2: Unfortunately, the authors could not add any data because of data availability of their database. That suggests this study has numbers of significant limitations.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Hiroo Kawarazaki

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files to be viewed.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org. Please note that Supporting Information files do not need this step.

27 Feb 2020

To Reviewer #1:

Dear Doctor Kawarazaki,

Thank you very much for your comment (as addressed below, and your positive evaluation. We really appreciate your time invested.

Best regards

Laurent Haas

1) Just one minor comment. Table A4 has systolic BP and diastolic BP the other way round. All comments have been addressed.

Thank you for your perceptiveness. We have corrected the mistake and also some minor formatting issues in our tables.

Reviewer #2:

Dear Doctor Kent Doi,

Thank you for your work on our manuscript. In your comment you pointed to the limitations of the study and particularly the missing information. With this in mind, we returned to your comments of the first review. In this second revision we further addressed your comments as follows.

1) Were end-stage renal disease patients treated by maintenance dialysis included? If so, their eGFR should be different between before and after dialysis. This will cause a serious bias.

The text was extensively revised. Excluding patients with an eGFR below 15 ml/min (Supplemental Material Table A3) did not change the results significantly. The limitation was included in the Abstract and Discussion, respectively.

2) Cat 4 (eGFR 15-29) showed worse survival curve just before day 30 compared with Cat 5 (eGFR<15) (Figure 2). The reason for this should be investigated. Did Cat 5 include many dialysis patients?

This remains as an intriguing finding. The number of patients and the event rate is small in patients with a eGFR<15 ml/min. Only three events would have lead to the same event rate as in the group with a eGFR 15-30 ml/min. Therefore, the group size might be of importance. Furthermore, 30-day mortality might be the better parameter as discussed.

3) It is a little bit surprising that most (72.8%) pts were treated as inpatients in this ED cohort. Can the severity of the enrolled patients be evaluated? Can qSOFA score be calculated?

Our ED consist of two wards. One is managed by general physicians and foreseen for patients likely to be treated as outpatients. In the other ward, patients likely to be treated as inpatients are assigned to. The study enrolled only patients in the latter ward, which explains the high rate of inpatients.

qSOFA cannot be calculated as the information is not available. The qSOFA would describe the severity in septic patients. Only, 15% suffered from infectious diseases. Therefore, the gained information would be limited.

4) As mentioned in the limitation section, this study did not distinguish AKI from CKD. However, because 72% of the enrolled patients admitted, subanalysis regarding AKI, CKD, acute-on-chronic can be performed when limited to the inpatients. This analysis will provide more precise role of single measurement of eGFR at ED.

Based on the helpful suggestions of the reviewers, the text was extensively revised and focused on the answers which can be given with the current dataset.

Submitted filename: Response to Reviewers.docx

Click here for additional data file.

16 Mar 2020

Estimated glomerular filtration rate predicts 30-day mortality in medical emergency departments: results of a prospective multi-national observational study

PONE-D-19-23285R2

Dear Dr. Haas,

We are pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it complies with all outstanding technical requirements.

Within one week, you will receive an e-mail containing information on the amendments required prior to publication. When all required modifications have been addressed, you will receive a formal acceptance letter and your manuscript will proceed to our production department and be scheduled for publication.

Shortly after the formal acceptance letter is sent, an invoice for payment will follow. To ensure an efficient production and billing process, please log into Editorial Manager at https://www.editorialmanager.com/pone/, click the "Update My Information" link at the top of the page, and update your user information. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to enable them to help maximize its impact. If they will be preparing press materials for this manuscript, you must inform our press team as soon as possible and no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

With kind regards,

Tatsuo Shimosawa, M.D., Ph.D.

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

23 Mar 2020

PONE-D-19-23285R2

Estimated glomerular filtration rate predicts 30-day mortality in medical emergency departments: Results of a prospective multi-national observational study

Dear Dr. Haas:

I am pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximize its impact. If they will be preparing press materials for this manuscript, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

For any other questions or concerns, please email plosone@plos.org.

Thank you for submitting your work to PLOS ONE.

With kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Prof. Tatsuo Shimosawa

Academic Editor

PLOS ONE

Table 1

Baseline characteristics of the total cohort and stratified by eGFR categories.

		eGFR (ml/min/1.73m2)
	Total cohort	>90	60–89	45–59	30–44	15–29	<15
Number (%)	6983	2544 (36%)	2504 (35%)	889 (12%)	561 (9%)	309 (5%)	176 (3%)
Age(years; median,(IQR)	62 (46, 76)	44 (31, 56)	67 (55, 76)	77 (68, 84)	79 (69, 85)	79 (71, 85)	69 (58, 82)
Male Sex	3723 (53.3%)	1288 (50.6%)	1403 (56.0%)	463 (52.1%)	308 (54.9%)	162 (52.4%)	99 (56.2%)
Vital signs, median (IQR)
Systolic blood pressure (mmHg)	137 (121, 154)	133 (120, 148)	142 (126, 158)	141 (123, 160)	137.50 (117, 155)	127 (105, 148)	130 (107, 155)
Diastolic blood pressure (mmHg)	80 (70, 90)	82 (73, 91)	81 (72, 92)	78 (68, 90)	75 (65, 85)	69 (58, 81)	70 (56, 85)
Pulse rate (bpm)	83 (71, 97)	84 (73, 97)	82 (70, 96)	82 (70, 93)	82 (71, 99)	83 (70, 98)	83 (69, 98)
O2 Saturation (%)	96 (94, 98)	97 (96, 99)	96 (94, 98)	96 (93, 97)	95 (92, 97)	95 (93, 97.4)	96 (92, 98)
Temperature (°C)	36.8 (36.4, 37.2)	36.8 (36.5, 37.2)	36.8 (36.4, 37.2)	36.8 (36.4, 37.3)	36.8 (36.5, 37.5)	36.7 (36.3, 37.3)	36.7 (36.2, 37.2)
Laboratory results, median (IQR)
Hemoglobin (g/l),	13.6 (12.1, 14.8)	14 (12.8, 15.1)	13.8 (12.5, 14.9)	13.1 (11.6, 14.5)	12.3 (10.8, 13.8)	11.4 (9.8, 13.0)	10.9 (9.4, 12.4)
Leukocyte count (G/l)	8.37 (6.57, 10.96)	8.05 (6.36, 10.4)	8.32 (6.50, 10.71)	8.59 (6.80, 11.24)	9.29 (6.90, 12.59)	9.30 (6.84, 13.01)	9.26 (7.09, 13.20)
Glucose (mmol/l)	6.1 (5.4, 7.5)	5.7 (5.1, 6.5)	6.3 (5.5, 7.5)	6.6 (5.7, 8.2)	7 (5.8, 9.3)	7.1 (6.0, 9.2)	6.9 (5.3, 8.9)
Creatinine (μmol/l)	81 (67, 103)	64 (56, 74)	82 (72, 94)	106 (91, 117)	136 (112, 156)	209 (179, 243)	487 (369, 686)
eGFR (ml/min/1.73m2)	80 (57, 98)	104 (97, 115)	76 (68, 83)	53 (49, 56)	38 (34, 41)	23 (18, 26)	8 (6, 12)
CRP (mg/l)	5.4 (<3, 27.8)	1 (<3, 11.7)	4.8 (<3, 22.5)	9.3 (<3, 48.7)	18.1 (4.9, 74.7)	32.9 (7.4, 116)	32.3 (6.1, 159)
PCT (μg/l)	0.08 (0.06, 0.13)	0.07 (0.05, 0.10)	0.07 (0.06, 0.11)	0.09 (0.07, 0.14)	0.12 (0.08, 0.25)	0.21 (0.12, 0.51)	0.47 (0.21, 1.64)
Copeptin (pmol/l)	10.8 (4.5, 39.2)	5.8 (3.3, 12.5)	10.1 (4.7, 31.6)	22.6 (8.6, 61.0)	42.9 (17.5, 101.0)	85.6 (41.6, 161.1)	136 (59.7, 218.0)
Main diagnosis, number, (%)
Infection	1026 (14.7%)	333 (13.1%)	323 (12.9%)	145 (16.3%)	106 (18.9%)	81 (26.2%)	38 (21.6%)
Cardiovascular	1636 (23.4%)	512 (20.1%)	648 (25.9%)	226 (25.4%)	148 (26.4%)	62 (20.1%)	40 (22.7%)
Metabolic	190 (2.7%)	58 (2.3%)	44 (1.8%)	16 (1.8%)	24 (4.3%)	17 (5.5%)	31 (17.6%)
Cancer	339 (4.9%)	101 (4.0%)	126 (5.0%)	42 (4.7%)	45 (8.0%)	15 (4.9%)	10 (5.7%)
Neurological	1549 (22.2%)	575 (22.6%)	613 (24.5%)	210 (23.6%)	96 (17.1%)	42 (13.6%)	13 (7.4%)
Gastroinstestinal	977 (14.0%)	426 (16.7%)	301 (12.0%)	106 (11.9%)	67 (11.9%)	48 (15.5%)	29 (16.5%)
Pulmonary	292 (4.2%)	115 (4.5%)	104 (4.2%)	41 (4.6%)	20 (3.6%)	9 (2.9%)	3 (1.7%)
Other	974 (13.9%)	424 (16.7%)	345 (13.8%)	103 (11.6%)	55 (9.8%)	35 (11.3%)	12 (6.8%)
Comorbidities, number, (%)
Cancer	960 (13.7%)	251 (9.9%)	377 (15.1%)	145 (16.3%)	102 (18.2%)	57 (18.4%)	28 (15.9%)
Renal failure	864 (12.4%)	30 (1.2%)	56 (2.2%)	174 (19.6%)	264 (47.1%)	219 (70.9%)	121 (68.8%)
Congestive heart disease	481 (6.9%)	38 (1.5%)	133 (5.3%)	124 (13.9%)	98 (17.5%)	63 (20.4%)	25 (14.2%)
COPD	355 (5.1%)	95 (3.7%)	127 (5.1%)	60 (6.7%)	43 (7.7%)	25 (8.1%)	5 (2.8%)
Coronary heart disease	834 (11.9%)	146 (5.7%)	350 (14.0%)	153 (17.2%)	101 (18.0%)	55 (17.8%)	29 (16.5%)
Dementia	220 (3.2%)	12 (0.5%)	68 (2.7%)	76 (8.5%)	39 (7.0%)	14 (4.5%)	11 (6.2%)
Diabetes mellitus	1075 (15.4%)	183 (7.2%)	374 (14.9%)	191 (21.5%)	169 (30.1%)	96 (31.1%)	62 (35.2%)
History of Stroke	564 (8.1%)	124 (4.9%)	235 (9.4%)	118 (13.3%)	53 (9.4%)	30 (9.7%)	4 (2.3%)
Hypertension	2769 (39.7%)	486 (19.1%)	1138 (45.4%)	551 (62.0%)	335 (59.7%)	182 (58.9%)	77 (43.8%)

Table 2

Frequencies of primary and secondary endpoints according to eGFR groups and associations of eGFR stratified by groups with adverse clinical outcome in univariate and multivariate regression analyses.

		eGFR (ml/min/1.73m2)
	Total cohort	>90	60–89	45–59	30–44	15–29	<15
30-day mortality, number (%)	325/6983 (4.65%)	47/2544 (1.8%)	88/2504 (3.5%)	61/889 (6.9%)	62/561 (11.1%)	42/309 (13.6%)	25/176 (14.2%)
Odds ratio (95% CI), p-value
Unadjusted model		Ref	1.94 (1.35 to 2.77), p<0.001	3.91 (2.65 to 5.77), p<0.001	6.60 (4.46 to 9.76), p<0.001	8.36 (5.41 to 12.91), p<0.001	8.80 (5.27 to 14.68), p<0.001
Model 1		Ref	0.75 (0.50 to 1.12), p = 0.162	1.01 (0.64 to 1.60), p = 0.965	1.63 (1.02 to 2.60), p = 0.042	2.03 (1.22 to 3.38), p = 0.006	2.98 (1.70 to 5.23), p<0.001
Model 2		Ref	0.73 (0.49 to 1.1), p = 0.132	1.00 (0.63 to 1.58), p = 1	1.59 (0.99 to 2.54), p = 0.053	2.01 (1.21 to 3.33), p = 0.007	2.93 (1.67 to 5.13), p<0.001
Model 3		Ref	0.79 (0.53 to 1.19), p = 0.264	1.18 (0.74 to 1.88), p = 0.489	1.70 (1.05 to 2.75), p = 0.03	2.31 (1.36 to 3.90), p = 0.002	3.73 (2.04 to 6.84), p<0.001
In-hospital mortality, number (%)	183 (2.62%)	27 (1.1%)	49 (2.0%)	27 (3.0%)	41 (7.3%)	27 (8.7%)	12 (6.8%)
Odds ratio (95% CI), p-value
Unadjusted model		Ref	1.86 (1.16 to 2.99), p = 0.01	2.92 (1.70 to 5.01), p<0.001	7.35 (4.48 to 12.06), p<0.001	8.93 (5.16 to 15.43), p<0.001	6.82 (3.39 to 13.71), p<0.001
Model 1		Ref	0.68 (0.40 to 1.16), p = 0.162	0.70 (0.37 to 1.32), p = 0.268	1.68 (0.92 to 3.06), p = 0.092	2.01 (1.05 to 3.83), p = 0.035	2.12 (0.99 to 4.53), p = 0.054
Model 2		Ref	0.67 (0.39 to 1.14), p = 0.138	0.69 (0.37 to 1.30), p = 0.254	1.64 (0.90 to 2.99), p = 0.107	1.98 (1.04 to 3.78), p = 0.037	2.07 (0.97 to 4.43), p = 0.06
Model 3		Ref	0.75 (0.44 to 1.27), p = 0.287	0.80 (0.42 to 1.50), p = 0.479	1.69 (0.92 to 3.10), p = 0.088	2.11 (1.09 to 4.08), p = 0.026	2.53 (1.13 to 5.64), p = 0.024
ICU admission, number (%)	441 (6.32%)	98 (3.9%)	146 (5.8%)	81 (9.1%)	51 (9.1%)	34 (11.0%)	31 (17.6%)
Odds ratio (95% CI), p-value
Unadjusted model		Ref	1.55 (1.19 to 2.01), p = 0.001	2.50 (1.84 to 3.39), p<0.001	2.50 (1.76 to 3.55), p<0.001	3.09 (2.05 to 4.65), p<0.001	5.34 (3.45 to 8.26), p<0.001
Model 1		Ref	1.47 (1.10 to 1.98), p = 0.01	2.33 (1.61 to 3.36), p<0.001	2.32 (1.54 to 3.49), p<0.001	2.86 (1.80 to 4.55), p<0.001	5.05 (3.17 to 8.04), p<0.001
Model 2		Ref	1.45 (1.08 to 1.95), p = 0.014	2.31 (1.60 to 3.33), p<0.001	2.28 (1.52 to 3.44), p<0.001	2.84 (1.79 to 4.50), p<0.001	4.97 (3.12 to 7.91), p<0.001
Model 3		Ref	1.43 (1.06 to 1.93), p = 0.02	2.11 (1.45 to 3.08), p<0.001	2.10 (1.38 to 3.20), p = 0.001	2.55 (1.58 to 4.13), p<0.001	4.84 (2.96 to 7.90), p<0.001
Rehospitalisation, number (%)	581 (8.32%)	167 (6.56%)	223 (8.91%)	89 (10.01%)	51 (9.09%)	29 (9.39%)	22 (12.50%)
Odds ratio (95% CI), p-value
Unadjusted model		Ref	1.39 (1.13 to 1.71), p = 0.002	1.58 (1.21 to 2.07), p = 0.001	1.42 (1.03 to 1.98), p = 0.035	1.47 (0.98 to 2.23), p = 0.066	2.03 (1.27 to 3.26), p = 0.003
Model 1		Ref	1.28 (1.01 to 1.63), p = 0.043	1.40 (1.01 to 1.93), p = 0.041	1.25 (0.86 to 1.82), p = 0.238	1.29 (0.82 to 2.04), p = 0.266	1.85 (1.13 to 3.02), p = 0.015
Model 2		Ref	1.27 (1.00 to 1.61), p = 0.052	1.40 (1.01 to 1.93), p = 0.042	1.24 (0.86 to 1.81), p = 0.254	1.29 (0.82 to 2.03), p = 0.273	1.83 (1.12 to 2.99), p = 0.016
Model 3		Ref	1.25 (0.98 to 1.59), p = 0.067	1.35 (0.98 to 1.87), p = 0.07	1.14 (0.78 to 1.67), p = 0.49	1.17 (0.74 to 1.85), p = 0.508	1.51 (0.91 to 2.51), p = 0.114
LOS (days), mean (SD)	4.39 (5.82)	2.78 (5.13)	4.53 (5.87)	5.82 (5.82)	6.24 (6.46)	7.06 (5.20)	7.39 (6.26)
Unadjusted model		Ref	1.76 (1.41 to 2.11), p<0.001	3.05 (2.57 to 3.53), p<0.001	3.47 (2.89 to 4.04), p<0.001	4.29 (3.54 to 5.04), p<0.001	4.62 (3.67 to 5.56), p<0.001
Model 1		Ref	0.60 (0.20 to 1.00), p = 0.003	1.32 (0.77 to 1.88), p<0.001	1.73 (1.08 to 2.37), p<0.001	2.49 (1.68 to 3.29), p<0.001	3.40 (2.44 to 4.36), p<0.001
Model 2		Ref	0.58 (0.18 to 0.98), p = 0.004	1.34 (0.78 to 1.89), p<0.001	1.72 (1.07 to 2.36), p<0.001	2.50 (1.70 to 3.31), p<0.001	3.37 (2.41 to 4.33), p<0.001
Model 3		Ref	0.55 (0.16 to 0.94), p = 0.005	1.09 (0.55 to 1.64), p<0.001	1.27 (0.64 to 1.91), p<0.001	1.99 (1.20 to 2.78), p<0.001	3.16 (2.21 to 4.11), p<0.001

95% Confidence interval, eGFR estimated glomerular filtration rate, ICU intensive care unit, LOS length of stay, OR Odds ratio, models were stepwise adjusted for age (model 1), age, and gender (model 2), age, gender, main diagnosis, and comorbidities (model 3)

*Since LOS is a continuous variable regression results represent a regression coefficient (95% confidence interval)

Table 4

Discriminative performance of eGFR and serum creatinine for the prediction of the different outcomes.

	eGFR	Creatinine
	AUC
30-day mortality	0.71 (0.68 to 0.73)	0.65 (0.61 to 0.68)
In-hospital mortality	0.71 (0.68 to 0.75)	0.66 (0.61 to 0.70)
ICU admission	0.63 (0.60 to 0.65)	0.61 (0.58 to 0.64)

AUC area under the receiver operating curve, eGFR estimated glomerular filtration rate, ICU intensive care unit.