How long does it take for stage 3 ckd to progress to stage 4

Journal Article

Maneesh Sud,

1

Department of Medicine

,

University of Toronto

,

Toronto, Ontario

,

Canada

Search for other works by this author on:

Accepted:

17 October 2015

Published:

20 November 2015

• 
  PDF
• Split View
• • Article contents
  • Figures & tables
  • Video
  • Audio
  • Supplementary Data

• Cite

  Cite

  Maneesh Sud, Navdeep Tangri, Melania Pintilie, Andrew S. Levey, David M.J. Naimark, Progression to Stage 4 chronic kidney disease and death, acute kidney injury and hospitalization risk: a retrospective cohort study, Nephrology Dialysis Transplantation, Volume 31, Issue 7, July 2016, Pages 1122–1130, https://doi.org/10.1093/ndt/gfv389

  Close

• • Email
  • Twitter
  • Facebook
  • More

Close

Navbar Search Filter Microsite Search Term Search

Abstract

INTRODUCTION

Chronic kidney disease (CKD) affects up to 10–16% of the population [1, 2], and it is associated with an increased risk of end-stage renal disease (ESRD) and cardiovascular (CV) events [3, 4]. Stage 4 CKD, defined by a glomerular filtration rate (GFR) between 15 and 29 mL/min/1.73 m2, is associated with higher morbidity, mortality and costs compared with patients with earlier stages of CKD (i.e. Stage 3 CKD, GFR: 30–59 mL/min/1.73 m2) [3, 5–7], and current guidelines recommend greater intensity of care for patients with Stage 4 disease [3, 5, 6, 8–11]. However, there is little information on the risk of subsequent adverse outcomes for patients who progress from Stage 3 to Stage 4 CKD compared with otherwise similar patients who do not progress. Understanding these risks may have potential clinical, research and public health implications. For these reasons, we examined the association between progression to Stage 4 CKD and clinically meaningful outcomes including death, acute kidney injury (AKI) and all-cause hospitalization prior to ESRD.

MATERIALS AND METHODS

Study cohort

We analyzed the previously described Sunnybrook CKD cohort [12, 13], consisting of all outpatient referrals for patients with CKD to the nephrology clinic at Sunnybrook Health Science Centre (SHSC) prior to 31 December 2008 (n = 6255) [13]. Since hospitalization data were available between 1 January 2001 and 31 December 2008, patients referred prior to 1 January 2001 (n = 603) were excluded. Patients for whom an estimated GFR (eGFR) was not available at the time of referral (n = 561), had an eGFR ≥60 or <30 mL/min/1.73 m2 (n = 3089), lacked at least two eGFR measurements (n = 392) or were hospitalized, developed ESRD or died prior to their first clinic date (n = 3), were excluded in order to yield the final cohort of 1607 subjects with Stage 3 CKD (equivalent to category G3 of the new KDIGO classification [1]; Figure 1).

FIGURE 1:

Flow diagram. CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; ESRD, end-stage renal disease.

Definition and ascertainment of exposure and outcome variables

The origin for time-to-event analyses was defined as the date of the first nephrology clinic visit. The interim exposure variable of interest was progression to Stage 4 CKD (eGFR <30 mL/min/1.73 m2) during the follow-up period (1 January 2001 to 31 December 2008) defined by two or more consecutive eGFR values <30 mL/min/1.73 m2 with the date of the second reading taken as date of progression to Stage 4 CKD. In comparison with the number of patients who progressed to Stage 4 (Stage 4 progressions), we determined the number of patients with an eGFR decline of 57% or more from baseline on two or more consecutive measurements. An eGFR decline of 57% is equivalent to a doubling of serum creatinine, which is an accepted surrogate outcome in clinical trials.

Baseline covariates included demographic variables—age and sex; physical examination data—blood pressure and comorbid conditions—diabetes, hypertension and heart failure, and occlusive CV disease (peripheral, coronary or cerebral vascular disease), which were defined as absent or present [13]. Baseline laboratory variables were defined as the first result closest to and within 365 days of the origin and at least 30 days prior to the first eGFR measurement <30 mL/min/1.73 m2. The source of baseline serum creatinine measurements is shown in Supplementary data, Table S1. GFR was estimated using the Chronic Kidney Disease Epidemiology (CKD-EPI) 2009 creatinine equation [14]. Albuminuria was defined either by log-transformed, measured urine albumin-to-creatinine ratio (ACR) or from 24-h urinary protein excretion transformed to an equivalent log-ACR [15]. Except for ACR, variables with >25% missing values were not included in the main analysis. All other missing data were imputed according to the multiple imputation technique [16] with 20 imputations as implemented in the PROC MI procedure in SAS version 9.3 using the default MCMC options (SAS Institute Inc., Cary, NC, USA).

The primary outcomes of interest included death, first AKI hospitalization and first all-cause hospitalization prior to ESRD. We considered AKI hospitalization to include the first admission with an AKI code as a primary or secondary diagnosis. We ascertained the first all-cause hospitalization over the follow-up period without restriction on the reason for hospitalization or length of stay. ESRD was defined as the need for chronic renal replacement therapy (RRT, dialysis or preemptive renal transplantation). In a secondary analysis, we partitioned all-cause hospitalization, whether associated with AKI or not, into those with or without a primary or secondary CV diagnostic code (myocardial infarction, heart failure and stroke), henceforth referred to as CV-related hospitalization and non-CV-related hospitalization, respectively.

Hospitalizations were ascertained via linkage to the Canadian Institute of Health Information Discharge Abstract Database that contains information on hospitalizations within Ontario including admission dates and diagnostic codes for up to eight diagnoses in the International Classification of Diseases (ICD) 9th revision, in 2001, and 25 diagnoses from the ICD 10th revision from 2002 to 2008. We used the date of admission for all hospitalizations in time-to-event analyses. ESRD was ascertained via linkage to the Ontario Ministry of Health Physician Claims Database that contains all records of physician billing for outpatient and inpatient services in Ontario including a service date and diagnosis for each claim. The date of ESRD initiation was taken as that for the first fee claim for chronic dialysis or preemptive renal transplantation. Death was ascertained from the Ontario Registered Persons Database, which that contains vital statistics on all residents of Ontario eligible for the Ontario Health Insurance Plan [17]. The sets of codes used to ascertain AKI, CV-related and other hospitalizations, subsequent ESRD and death prior to ESRD, as well as their diagnostic performance in validation studies, are listed in Supplementary data, Table S2.

Statistical analysis

For the main analysis, we developed Cox proportional hazards models to estimate cause-specific hazard ratios (csHRs) associated with a binary time-dependent covariate (progression to Stage 4 CKD). csHRs were estimated both with and without adjustment for baseline time-independent covariates (age, gender, blood pressure, history of diabetes, hypertension, occlusive CV disease, heart failure, cancer, chronic obstructive pulmonary disease and dementia, as well as eGFR, ACR, calcium, phosphate, bicarbonate, urea, albumin and hemoglobin) [8]. We developed three separate models: first, ESRD or death prior to ESRD; second, ESRD, death prior to ESRD or AKI hospitalization prior to ESRD; and third, ESRD, death prior to ESRD or all-cause hospitalization prior to ESRD [18]. In a secondary analysis, we developed models for ESRD, death prior to ESRD; and either CV or non-CV hospitalization as defined above. For the latter models, we tested for the significance of differences of the csHRs associated with interim progression to Stage 4 between the outcomes of CV and non-CV hospitalization according to the method of Lunn and McNeil [19]. Since all patients who developed ESRD during the follow-up period also progressed from Stage 3 to Stage 4 CKD, we did not estimate csHR of Stage 4 progression on this outcome. In supplemental analyses, we computed competing risk, cumulative incidence functions for the outcomes described above and we employed Fine and Gray regression models [20, 21] to estimate sub-distribution HRs (sdHRs) associated with a binary time-dependent covariate (progression to Stage 4 CKD) adjusted for the same baseline covariates and outcomes indicated above.

We conducted additional adjusted subgroup analyses, defined a priori, across baseline characteristics including age less than versus greater than or equal to the cohort median age (70 years), male versus female, preexisting diabetes versus no preexisting diabetes, preexisting hypertension versus no preexisting hypertension history and eGFR ≥45 mL/min/1.73 m2 versus <45 mL/min/1.73 m2 (i.e. 3A versus 3B CKD). Sensitivity adjusted analyses were also undertaken to assess the robustness of the results. Age as the time scale was used rather than time-on-study to rule out residual confounding of risk of each outcome by age at the time of origin that is not sufficiently accounted for by including baseline age as a predictor variable in the proportional hazards model [22]. The analysis was repeated with complete cases only (n = 958). Lastly, we modified the cohort by redefining progression to Stage 4 CKD as either three (n = 1362) or five (n = 1362) follow-up eGFR values <30 mL/min/1.73 m2.

For baseline covariates, which were assumed to be time independent in the main Cox models, where there was evidence of non-proportionality and the effect of the coefficient changed with time, time-averaged effects were reported in the main analysis [23]. For the latter factors, supplemental analyses were performed, which included covariate-by-time interaction terms. Except for data imputation (above), all statistical analyses were carried out using open source R v3.0.2 software. A two-sided significance level of P < 0.05 was employed for all analyses. This study was approved by the Research Ethics Board at SHSC, and a waiver of consent was granted given its retrospective nature.

RESULTS

Study cohort characteristics and event rates

The 1607 subjects with Stage 3 CKD had a mean age of 69 years and mean [standard deviation (SD)] eGFR of 43 (8) mL/min/1.73 m2 with baseline characteristics and comorbid illnesses presented in Table 1. Over a median follow-up time of 2.66 years (interquartile range: 1.42–4.45), there were 344 (21%) patients who progressed to Stage 4 CKD, 68 (4%) who developed a 57% decline in eGFR and 47 (3%) who developed ESRD. The number of available creatinine (eGFR) values and the time between measurements were similar between subjects who did and did not progress (Supplementary data, Figures S1 and S2). Among the 344 patients who progressed to Stage 4 CKD, 57 (17%) were Stage 3A (45–59 mL/min/1.73 m2) and 287 (83%) were Stage 3B (30–44 mL/min/1.73 m2) at baseline. At the time of last follow-up, the mean (SD) eGFR in patients who progressed to Stage 4 CKD was 24 (12) compared with 46 (16) mL/min/1.73 m2 in those who did not progress. During the same time period, 188 (12%) patients died, 143 (9%) were hospitalized with AKI and 688 (43%) were hospitalized for any reason prior to ESRD. The unadjusted rates per 100 patient-years of death, AKI hospitalization and all-cause hospitalization prior to ESRD were higher in patients who progressed to Stage 4 CKD when compared with patients who did not progress (Figure 2).

Table 1.

Baseline characteristics and outcomes

N, n, number; SD, standard deviation; IQR, interquartile range; eGFR, estimated glomerular filtration rate; Stage 4, eGFR <30 mL/min/1.73 m2, AKI, acute kidney injury; ESRD, end-stage renal disease.

aCancer of any origin including skin cancers.

bIn an adjusted, cause-specific model, two outcomes were considered: ESRD and deaths prior to ESRD during the follow-up period.

cBy definition, participants who did not progress to Stage 4 could not have progressed to ESRD or developed a 57% reduction in baseline eGFR.

dA 57% reduction in eGFR corresponds to approximately a doubling of the serum creatinine concentration. Two or more eGFR values that were at least 57% lower than the baseline value were required to classify participants as having suffered a 57% eGFR reduction in order to be comparable to the requirement of two eGFR values ≤30 mL/min/1.73 m2 in order to be classified as having progressed to Stage 4 CKD.

eAKI hospitalization outcomes were analyzed in an adjusted, cause-specific model that included ESRD and death prior to ESRD as alternative outcomes during the follow-up period.

fAll-cause hospitalization outcomes were analyzed in an adjusted, cause-specific model that included ESRD and death prior to ESRD as alternative outcomes during the follow-up period.

Table 1.

Baseline characteristics and outcomes

N, n, number; SD, standard deviation; IQR, interquartile range; eGFR, estimated glomerular filtration rate; Stage 4, eGFR <30 mL/min/1.73 m2, AKI, acute kidney injury; ESRD, end-stage renal disease.

aCancer of any origin including skin cancers.

bIn an adjusted, cause-specific model, two outcomes were considered: ESRD and deaths prior to ESRD during the follow-up period.

cBy definition, participants who did not progress to Stage 4 could not have progressed to ESRD or developed a 57% reduction in baseline eGFR.

dA 57% reduction in eGFR corresponds to approximately a doubling of the serum creatinine concentration. Two or more eGFR values that were at least 57% lower than the baseline value were required to classify participants as having suffered a 57% eGFR reduction in order to be comparable to the requirement of two eGFR values ≤30 mL/min/1.73 m2 in order to be classified as having progressed to Stage 4 CKD.

eAKI hospitalization outcomes were analyzed in an adjusted, cause-specific model that included ESRD and death prior to ESRD as alternative outcomes during the follow-up period.

fAll-cause hospitalization outcomes were analyzed in an adjusted, cause-specific model that included ESRD and death prior to ESRD as alternative outcomes during the follow-up period.

FIGURE 2:

Rates of mortality and hospitalization. The rate per 100 patient-years in all patients, those who progressed to Stage 4 CKD and those who did not for each outcome prior to ESRD is depicted. AKI, acute kidney injury; ESRD, end-stage renal disease.

Association between progression to Stage 4 CKD and subsequent mortality and hospitalization

When compared with patients who did not progress to Stage 4 CKD over the follow-up period, those who progressed had significantly higher unadjusted risk of death [csHR = 3.03, 95% confidence interval (95% CI): 2.20–4.17], AKI hospitalization (csHR = 3.69, 95% CI: 2.50–5.44) and all-cause hospitalization (csHR = 2.21, 95% CI: 1.75–2.78) prior to ESRD. After adjustment for baseline predictor variables, including eGFR, the risk estimates were attenuated but remained significant for death (csHR = 2.56, 95% CI: 1.75–3.75), AKI hospitalization (csHR = 2.32, 95% CI: 1.44–3.74) and all-cause hospitalization (csHR = 1.87, 95% CI: 1.45–2.42) prior to ESRD (Figure 3). There was no statistically significant difference in the increased risk of death, AKI hospitalization and all-cause hospitalization prior to ESRD after progression to Stage 4 CKD across predefined subgroups (all P > 0.05 for interaction) (Figure 4).

FIGURE 3:

Risk of mortality and hospitalization after progression to Stage 4 CKD. The unadjusted and adjusted risk for each outcome prior to ESRD in patients who progressed to Stage 4 CKD when compared with patients who did not progress over the follow-up time is depicted. HR, hazard ratio; CI, confidence interval; AKI, acute kidney injury; ESRD, end-stage renal disease.

FIGURE 4:

Subgroup analysis. The adjusted risk for each outcome prior to ESRD in patients who progressed to Stage 4 CKD when compared with patients who did not progress across predefined subgroups is depicted. For all subgroups, test for interaction was nonsignificant (P > 0.05). HR, hazard ratio; CI, confidence interval; AKI, acute kidney injury; ESRD, end-stage renal disease.

A CV-related hospitalization accounted for 20% (n = 138) of the first hospitalization events during the follow-up period. When compared with patients who did not progress to Stage 4 CKD over the follow-up period, those who progressed had a significantly higher adjusted HR of both CV-related hospitalization (csHR = 3.00, 95% CI: 1.77–5.08) and non-CV-related hospitalization (csHR = 1.62, 95% CI: 1.21–2.16; P = 0.041 for the difference in csHRs; Figure 5). Cumulative incidence functions for the competing risk models are shown in Supplementary data, Figure S3. With Fine and Gray competing risk regression, sdHR values were obtained that were similar to the csHRs above (Supplementary data, Figure S4).

FIGURE 5:

CV and non-CV hospitalization after progression to Stage 4 CKD. The adjusted risk for all-cause hospitalization prior to ESRD in patients who progressed to Stage 4 CKD when compared with patients who did not progress was partitioned into hospitalizations for a CV- and non-CV-related cause and associated P-value for difference.

Sensitivity analyses

Results in the same direction as the main analysis were obtained using (i) age as the time scale rather than time-on-study [5], (ii) complete cases (n = 958), (iii) redefining the cohort as patients with at least five follow-up eGFRs (n = 1059) and progression to Stage 4 CKD as two consecutive clinic visits with eGFR <30 mL/min/1.73 m2 (n = 224) or (iv) redefining the cohort as three follow-up eGFRs (n = 1362) and progression to Stage 4 CKD as three consecutive clinic visits with eGFR <30 mL/min/1.73 m2 (n = 251) (Supplementary data, Table S3). A comparison of baseline characteristics between the included cohort and the 392 patients that did not have at least two follow-up eGFRs reveals that these excluded patients were older, less likely to be male or have comorbid diabetes, more likely to have more co-morbid obstructive lung disease and dementia, and had higher baseline eGFRs and lower ACR values (Supplementary data, Table S4). None of these excluded patients developed ESRD in the province of Ontario, 64 (17%) died, 27 (7%) were hospitalized with AKI and 132 (35%) were hospitalized for any cause prior to ESRD during the same time period. Some baseline covariates demonstrated departure from proportionality but inclusion of covariate-by-time interaction terms for these produced results similar to the main analysis (Supplementary data, Table S5).

DISCUSSION

In this cohort of patients followed in a nephrology clinic referred with Stage 3 CKD, 21% progressed to Stage 4 CKD over a median follow-up time of 3 years, whereas only 3% of patients developed ESRD and 4% developed a 57% reduction from baseline eGFR, equivalent to a doubling of baseline serum creatinine. We observed that progression to Stage 4 CKD was associated with a nearly 2-fold higher risk of mortality, AKI and all-cause hospitalizations when compared with patients who did not progress by the end of follow-up, that these risks were similar in subgroups of patients with either Stage 3A or 3B CKD at the origin and, consistent with recent literature concerning recurrent AKI hospitalization [24], that risk estimates were similar whether cause-specific or competing risk survival analyses were employed. Furthermore, we observed that the risk of CV-related hospitalizations after progression to Stage 4 was higher than non-CV-related hospitalizations. We analyzed mortality and hospitalization events that occurred prior to developing ESRD, thus demonstrating that the observed risks of mortality and morbidity were due to progression to Stage 4 CKD itself rather than as a result of subsequent ESRD. Finally, the risk of subsequent mortality and hospitalization remained significant after adjustment for baseline eGFR, age and other potential confounders, suggesting that progression to Stage 4 is independently associated with long-term adverse clinical outcomes.

To our knowledge, this is the first demonstration of the association between progression from Stage 3 to Stage 4 CKD and subsequent mortality and hospitalization prior to ESRD. Recent studies have demonstrated higher rates of mortality, CV events, hospitalizations, AKI and ESRD for patients with Stage 4 compared with Stage 3A and 3B patients [25, 26]. However, direct comparison of Stage 3 and 4 CKD patients may be biased since patients presenting with Stage 4 CKD would have had to endure a longer duration of CKD and consequently, a longer period of elevated mortality risk, compared with patients with Stage 3 CKD (i.e. a survivor bias). In the current analysis, we circumvented this issue by treating the progression to Stage 4 CKD as a time-dependent variable, comparing hospitalization and mortality risks subsequent to the progression, for those who progressed to Stage 4, with otherwise similar patients who did not progress to Stage 4 during the observation period. Our findings, therefore, suggest that the higher risk is related to the progression of CKD per se.

Our findings have important clinical implications. Progression to Stage 4 CKD represents a transition in the natural history of CKD when the burden of comorbid illnesses, particularly CV disease, rises [27], hormonal and metabolic perturbations, such as metabolic bone disease and anemia, become important [27] and complex decisions regarding RRT planning commence, all in the setting of increasing risks of ESRD and death [12]. We demonstrate that progression from Stage 3 to Stage 4 CKD identifies patients at especially high risk who require extra vigilance in the provision of guideline-recommended care.

Our findings may also have public health implications. Current KDIGO practice guidelines recommend several changes in the management of patients when they cross below the eGFR threshold of 30 mL/min/1.73 m2: for example, all such patients should be referred for nephrology specialty care [11], the frequency of laboratory monitoring should increase [10], vaccinations should be provided [8] and some medication changes may be required [8]. There may be patients with Stage 3 CKD, at higher risk of progression to Stage 4 CKD, who might benefit from the earlier introduction of these measures in order to reduce the subsequent risk of mortality and hospitalization observed in the current analysis. However, indiscriminate application of Stage 4 guidelines to Stage 3 patients would not be feasible since more intense management of Stage 4 CKD results in annual, incremental, per capita medical costs that are almost 4-fold greater than those associated with Stage 3 disease [7] and the proportion of the population with Stage 3 CKD is 15 times as large as that for Stage 4 CKD [1]. Without the ability to identify patients with Stage 3 CKD at high risk of progression to Stage 4, surveillance programs to assess adherence to guideline-recommended care for Stage 4 CKD are likely to remain more efficient than for Stage 3 CKD.

Our findings may also have implications for research in CKD. First, our results suggest that future development and validation of instruments to predict the subset of patients with Stage 3 most likely to progress to Stage 4 may have clinical value and be helpful for more efficient resource allocation. Instruments to predict the risk to ESRD are available [13], but the utility of these instruments to predict risk of progression to CKD Stage 4 has not been evaluated. Second, progression from Stage 3 to Stage 4 CKD could serve as a surrogate outcome for clinical trials. Currently, the US Food and Drug Administration accepts doubling of serum creatinine, equivalent to a 57% reduction in eGFR, as a surrogate outcome for kidney disease progression in clinical trials [28]. We observed an almost 5-fold higher number of progressions to Stage 4 than 57% eGFR declines, which if implemented as an alternative surrogate endpoint could potentially shorten the duration of clinical trials in patients with Stage 3 CKD. A recent meta-analysis of CKD cohorts, including the Sunnybrook CKD cohort, has demonstrated lesser declines of eGFR such as 30% associate with ESRD and mortality risk, supporting the role for alternative and less stringent surrogate outcomes [1]. Further studies are warranted to assess the usefulness of progression from Stage 3 to Stage 4 CKD as a surrogate outcome for clinical trials perhaps as part of a composite of progression to Stage 4 or 30% reduction in eGFR.

Our analysis has several limitations including those generally associated with administrative data. Information on treatment and race was lacking and one-third of the participants did not have albuminuria values, all of which may have affected risk estimates for both Stage 4 progression and outcomes. We did not have prospective, regularly scheduled, follow-up eGFR measurements for all participants, and we could not time the precise transition to Stage 4, which may affect the accuracy of the risk estimates. Additionally, we cannot be certain that a proportion of the analyzed cohort may have progressed to Stage 4 CKD, yet were not captured because follow-up eGFR values below 30 mL/min/1.73 m2 occurred after individuals had left the clinic or after the end of the study period. However, misclassification of patients who actually progressed to Stage 4 CKD as nonprogressors would only have attenuated the observed effects and, with proper reclassification, the effect sizes would likely have been larger. Our results that are derived from the experience of a single referral center may not be generalizable to cohorts in which referral practices differ from our own, notably when patients with Stage 3A CKD are managed in primary care without ongoing nephrology care. Furthermore, baseline laboratory variables for some patients were available after the origin. However, with respect to both of these issues, our previous work with the Kidney Failure Risk Equation was derived in the same population and has demonstrated excellent internal [13] and external [29] validity. Administrative codes for AKI have been shown to be insensitive compared with a laboratory-based gold standard [30]. However, such codes tend to select for a more severe, and therefore clinically relevant, AKI phenotype. Finally, we cannot be certain that the adverse effects observed in association with progression to Stage 4 CKD were not due to the effects of unmeasured confounders that worsened as kidney function declined.

In conclusion, we report, in a cohort of Stage 3 CKD patients, that progression to Stage 4 CKD confers a 2-fold increase in death, AKI hospitalization and all-cause hospitalization risk prior to ESRD when compared with not progressing. Further investigations are warranted to validate this finding and to eventually evaluate progression to Stage 4 CKD as a surrogate outcome for clinical trials.

AUTHORS’ CONTRIBUTIONS

M.S. and D.M.J.N. had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis [12]. Study concept and design: M.S., N.T., A.S.L. and D.M.J.N. Acquisition of data: M.S., N.T. and D.M.J.N. Statistical analysis: M.S. and M.P. Interpretation of data analysis: M.S., N.T., A.S.L. and D.M.J.N. Drafting of the manuscript: M.S. and D.M.J.N. Critical revision of the manuscript for important intellectual content: M.S., N.T., M.P., A.S.L. and D.M.J.N.

CONFLICT OF INTEREST STATEMENT

The authors have no conflicts of interest to declare and no financial disclosures to make with respect to this work. The results presented in this paper have not been published previously in whole or part, except in abstract format.

(See related article by Warnock. Competing risks: you only die once. Nephrol Dial Transplant 2016; 31: 1033–1035)

ACKNOWLEDGEMENTS

The authors wish to acknowledge Mr Alex Firby, AAPP, Senior Information Management Advisor, Negotiations Information Management Strategy & Policy Branch, and Terry Stevens, Coordinator, Health Data & Decision Support Unit, in the Ministry of Health and Long-Term Care, Province of Ontario, for their invaluable help in obtaining provincial data pertaining to the Sunnybrook cohort. The authors also wish to gratefully acknowledge the contribution of Dr Arthur Allignol of the University of Freiburg in Germany, for his assistance with the R code used in our time-dependent survival models. No compensation was received.

REFERENCES

1

KDIGO Working Group

.

KDIGO clinical practice guideline for the evaluation and management of chronic kidney disease. Chapter 1: definition and classification of CKD

. Kidney Int Suppl

2013

;

3

:

44

2

Levey

AS

Chronic kidney disease as a global public health problem: approaches and initiatives—a position statement from Kidney Disease Improving Global Outcomes

.

Kidney Int

2007

;

72

:

247

–

259

3

Astor

BC

Lower estimated glomerular filtration rate and higher albuminuria are associated with mortality and end-stage renal disease. A collaborative meta-analysis of kidney disease population cohorts

.

Kidney Int

2011

;

79

:

1331

–

1340

4

Tonelli

M

Chronic kidney disease and mortality risk: a systematic review

.

J Am Soc Nephrol

2006

;

17

:

2034

–

2047

5

van der Velde

M

Lower estimated glomerular filtration rate and higher albuminuria are associated with all-cause and cardiovascular mortality. A collaborative meta-analysis of high-risk population cohorts

.

Kidney Int

2011

;

79

:

1341

–

1352

6

Gansevoort

RT

Lower estimated GFR and higher albuminuria are associated with adverse kidney outcomes. A collaborative meta-analysis of general and high-risk population cohorts

.

Kidney Int

2011

;

80

:

93

–

104

7

Honeycutt

AA

Medical costs of CKD in the Medicare population

.

J Am Soc Nephrol

2013

;

24

:

1478

–

1483

8

KDIGO Working Group

.

KDIGO clinical practice guideline for the evaluation and management of chronic kidney disease. Chapter 4: other complications of CKD: CVD, medication dosing, patient safety, infections, hospitalizations, and caveates for investigating complications of CKD

.

Kidney Int Suppl

2013

;

3

:

21

9

KDIGO Working Group

.

KDIGO clinical practice guideline for the evaluation and management of chronic kidney disease. Chapter 3: management of progression and complications of CKD

.

Kidney Int Suppl

2013

;

3

:

18

10

KDIGO Working Group

.

KDIGO clinical practice guideline for the evaluation and management of chronic kidney disease. Chapter 2: definition, identification, and prediction of CKD progression

.

Kidney Int Suppl

2013

;

3

:

63

–

72

11

KDIGO Working Group

.

KDIGO clinical practice guideline for the evaluation and management of chronic kidney disease. Chapter 5: referral to specialists and models of care

.

Kidney Int Suppl

2013

;

3

:

112

–

119

12

Sud

M

CKD stage at nephrology referral and factors influencing the risks of ESRD and death

.

Am J Kidney Dis

2014

;

63

:

928

–

936

.

13

Tangri

N

A predictive model for progression of chronic kidney disease to kidney failure

.

JAMA

2011

;

305

:

1553

–

1559

14

Levey

AS

A new equation to estimate glomerular filtration rate

.

Ann Intern Med

2009

;

150

:

604

–

612

15

Parving

HH

The effect of irbesartan on the development of diabetic nephropathy in patients with type 2 diabetes

.

N Engl J Med

2001

;

345

:

870

–

878

16

Donders

AR

Review: a gentle introduction to imputation of missing values

.

J Clin Epidemiol

2006

;

59

:

1087

–

1091

17

Iron

K

Living and dying in Ontario: an opportunity for improved health information

.

ICES investigative report

.

Toronto

:

Institute for Clinical Evaluative Sciences

,

2008

18

Wolkewitz

M

Risk factors for the development of nosocomial pneumonia and mortality on intensive care units: application of competing risks models

.

Crit Care

2008

;

12

:

R44

19

Lunn

M

Applying Cox regression to competing risks

.

Biometrics

1995

;

51

:

524

–

532

20

Fine

JP

A proportional hazards model for the sub-distribution of a competing risk

.

J Am Stat Assoc

1999

;

94

:

496

–

509

21

Geskus

RB

Cause-specific cumulative incidence estimation and the fine and gray model under both left truncation and right censoring

.

Biometrics

2011

;

67

:

39

–

49

22

Thiebaut

AC

Choice of time-scale in Cox's model analysis of epidemiologic cohort data: a simulation study

.

Stat Med

2004

;

23

:

3803

–

3820

23

Latouche

A

A competing risks analysis should report results on all cause-specific hazards and cumulative incidence functions

.

J Clin Epidemiol

2013

;

66

:

648

–

653

24

Siew

ED

Predictors of recurrent AKI

.

J Am Soc Nephrol

2015

;

.

25

Go

AS

Chronic kidney disease and the risks of death, cardiovascular events, and hospitalization

.

[Erratum appears in N Engl J Med 2008; 18: 4]

. N Engl J Med

2004

;

351

:

1296

–

1305

26

Levey

AS

The definition, classification, and prognosis of chronic kidney disease: a KDIGO Controversies Conference report

.

[Erratum appears in Kidney Int 2011; 80: 1000]

.

Kidney Int

2011

;

80

:

17

–

28

27

Abboud

H

Clinical practice. Stage IV chronic kidney disease

.

N Engl J Med

2010

;

362

:

56

–

65

28

Stevens

LA

Surrogate end points for clinical trials of kidney disease progression

.

Clin J Am Soc Nephrol

2006

;

1

:

874

–

884

29

Peeters

MJ

Validation of the kidney failure risk equation in European CKD patients

.

Nephrol Dial Transplant

2013

;

28

:

1773

–

1779

30

Grams

ME

Performance and limitations of administrative data in the identification of AKI

.

Clin J Am Soc Nephrol

2014

;

9

:

682

–

689

© The Author 2015. Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved.

© The Author 2015. Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved.

Supplementary data

Comments

2 Comments

Reply to Dr. Acosta-Ochoa

17 December 2015

David M.J. Naimark (with Maneesh Sud, Navdeep Tangri, Melania Pintilie, Andrew S. Levy)

Asst. Prof., University of Toronto

We thank the correspondent for her insightful questions. We lacked reliable data regarding the etiology of kidney disease in our cohort. However, with the exception of glomerular and polycystic kidney disease, it is not uncommon for clinicians to lack certainty regarding the etiology of chronic kidney disease (CKD) given that renal biopsies are not without risk and are usually performed in a minority of CKD patients. In our cohort, only 10 (2.9%) and 15 (1.2%) of the subjects who did or did not progress to stage 4, respectively, underwent a renal biopsy at our institution during the study period. With respect to glomerulonephritis, only 18 (5.2%) and 56 (4.4%), had a diagnosis of either systemic lupus erthythematosus or glomerulonephritis, respectively, thus it would be unlikely if the complications of immunosuppressive therapy would account for our findings. In terms of the differentiation between acute kidney injury (AKI)-on-CKD and CKD progression, the fact that we required at least two consecutive out-patient estimated glomerular filtration rate (eGFR) values below 30 ml/min/1.73m2 to count as stage 4 progression reduced the likelihood of misclassifying an episode of AKI (with subsequent recovery) as progression to stage 4. Furthermore, in a sensitivity analyses, requiring three consecutive eGFR values below 30 ml/min/1.73m2 did not meaningfully change our results. As for the differential burden of comorbidity in subjects to who did and did not progress to stage 4, as we show in Supplemental Table S5 in our paper, we adjusted for a large number of baseline covariates including the presence of diabetes, congestive heart failure, occlusive cardiovascular disease, and, notably, baseline eGFR. Of course, residual confounding can never be ruled out with certainty in an observational study. Finally, with respect to the differential numbers of 3A and 3B CKD subjects among subjects who did and did not progress to stage 4, it is correct that there were a greater proportion of stage 3B subjects among those who progressed. However, we think that the key point is that the risks of stage 4 progression on subsequent death and all-cause hospitalization in our study had the same magnitude for stage 3A and 3B subjects. For AKI-related hospitalization, the risk associated with stage 4 CKD progression was actually greater for 3B subjects. These results would not be expected if mere proximity to the 30 ml/min/1.73m2 boundary was the chief driver of our findings. In order to cross into stage 4 CKD, by definition, the eGFR for subjects with stage 3A disease at baseline had to decline to a greater degree than subjects who started with stage 3B disease. If the magnitude of eGFR decline was the most important factor, we would expect the effect of stage 4 progression on hospitalization and death to be greater for stage 3A than for stage 3B subjects. This is not what we see in our analyses. We believe these observations actually bolster the contention that the fact of progression to stage 4 is more important rather than the size of the eGFR decline per se. In other words, an eGFR of 30 ml/min/1.73 m2 marks a significant pathophysiologic boundary.

Conflict of Interest:

None declared

Submitted on 17/12/2015 7:00 PM GMT

In regard to Sud et al.

17 December 2015

Isabel Acosta-Ochoa

Nephrologist, Hospital Clinico Universitario, Valladolid, Spain

We read Sud et al.'s paper with great interest and welcome the concept of CKD progression and its relation with AKI, CV events, all cause hospitalizations and death.

But we observe a paucity in data on CKD etiology and treatment (glomerulonephritis under immunosuppressive treatment would be a risk factor for sepsis, AKI, CV events and all cause hospitalization).

Another issue is how to differentiate real AKI-on-CKD and CKD progression, which is not mentioned in the paper, and the definition and degree of renal recovery (1,2). Progressors have more diabetes, heart failure and occlusive cardiovascular disease. More "no progressors" were in Stage 3A, whilst more "progressors" were in Stage 3B. We think that both groups are essentially different and that progressor bear more comorbidities (e.g. are more prone to be exposed to invasive procedures and nephrotoxins) and are closer to Stage 4.

We think that patients with CKD need individual care and therapies, and there is still much work to do in the field of prognostication (3), with classic and emergent risk factors entering in the equation. In the future novel biomarkers should help to differentiate between AKI-on-CKD and CKD progression (4).

References

1. Goldstein SL, Chawla L, Ronco C, Kellum JA. Renal recovery. Crit Care. 2014 Jan 6;18(2):301.

2. A. Sharma, MJ. Mucino, C. Ronco. Renal Functional Reserve and Renal Recovery after Acute Kidney Injury. Nephron Clin Pract 2014;127:94- 100.

3. Onuigbo MA, Agbasi N. Chronic kidney disease prediction is an inexact science: The concept of "progressors" and "nonprogressors". World J Nephrol. 2014 Aug 6;3(3):31-49.

4. ME.Wasung, LS. Chawla, M. Madero. Biomarkers of renal function, which and when? Clinica Chimica Acta 438 (2015) 350-357.

Conflict of Interest:

None declared

Submitted on 17/12/2015 7:00 PM GMT

I agree to the terms and conditions. You must accept the terms and conditions.

Submit a comment

Name

Affiliations

Comment title

Comment

You have entered an invalid code

Thank you for submitting a comment on this article. Your comment will be reviewed and published at the journal's discretion. Please check for further notifications by email.

Citations

Views

Altmetric

Email alerts

More on this topic

Related articles in PubMed

Citing articles via

• Latest

• Most Read

• Most Cited

More from Oxford Academic

Does Stage 3 kidney failure always progress to Stage 4?

How long can you stay in stage 3 kidney disease life expectancy?

Should I be worried about CKD Stage 3?

How quickly does renal failure progress?