Automated laboratory-based reporting of eGFR has been widely adopted globally in the last decade to aid in the early recognition of CKD.1–4 There remains, however, limited knowledge regarding the impact of this public health intervention on patient outcomes and health care resource use. Several studies have shown that, although routine eGFR reporting seems to result in an increase in appropriate referral for specialist care and medication prescription, it may also increase the number of unnecessary referrals and the number of patients inappropriately labeled with a chronic disease.1,5–8 Furthermore, with eGFR reporting, the identification of kidney disease is not uniform, because older individuals with normal creatinine values are disproportionately reclassified as having CKD and subsequently referred for specialist care.1,8,9
In parallel with the introduction of automated eGFR reporting, a steady rise in the eGFR at dialysis initiation has been observed over the same decade.10–12 This trend is concerning, because the initiation of dialysis at higher levels of eGFR is associated with early mortality and excess costs.11,13–15 Reasons for why these trends are occurring seem multifactorial.16–18 Dialysis patients are, on average, older with substantial comorbid illnesses; therefore, determining whether nonspecific symptoms, such as fatigue, are attributed to uremia or underlying illness is difficult. Dialysis initiation at higher eGFR levels seems to occur more commonly in elderly individuals, the identical cohort with an increase in nephrology referrals because of eGFR reporting.
From a clinical and health policy perspective, it is important to understand whether automated eGFR reporting has contributed to earlier dialysis initiation. Thus, we hypothesized that automated eGFR reporting may be leading to earlier dialysis initiation based on eGFR level. The objectives of this study were, therefore, to examine the association of eGFR reporting and the change in eGFR level at dialysis initiation over time among incident dialysis patient populations.
Results
Characteristics Pre- and Post-eGFR Reporting
During the study period, 23,674 patients initiated dialysis, of whom 22,208 (93.4%) had serum creatinine measure at dialysis initiation; 9666 and 12,542 patients initiated hemodialysis in the provinces of interest during the pre- and post-eGFR reporting periods, respectively. Compared with the prereporting interval, patients in the postreporting interval were of similar age, more commonly men, and had a higher body mass index. In the postreporting interval, malignancy, coronary artery bypass grafting, and any serious illness were more common, whereas cardiovascular disease (excluding stroke), pulmonary edema, and hypertensive medication usage decreased (Table 1). Hypertension as a cause of ESRD was more frequent in the recent era, whereas the average hemoglobin and serum phosphate decreased. Use of an arteriovenous fistula (AVF) for hemodialysis and use of peritoneal dialysis both declined over time.
;)
Table 1: Baseline characteristics of incident dialysis patients pre- and post-eGFR reporting
Changes in the eGFR Level at Dialysis Initiation after eGFR Reporting
The mean eGFR and proportion of patient populations with eGFR≥10.5 and ≥12 ml/min per m2 at dialysis initiation increased after eGFR reporting was implemented (Table 2); however, in the unadjusted interrupted time series, the increase was linear, with no discernable change in trend because of eGFR reporting (Figure 1A) (P value for change in slope=0.66, P value for change in intercept=0.66). This finding was consistent when examining mean eGFR and proportion of patients with eGFR≥10.5 or ≥12 ml/min per m2 (Figure 1, B and C). Within age and sex strata, the mean eGFR was higher in the postreporting period in all categories except elderly men (P=0.26); however, in the unadjusted interrupted time series (Table 3), there were no significant changes in trend because of eGFR reporting (P values for change in slope were all nonsignificant).
Table 2: Mean and proportional changes in the eGFR level at dialysis initiation among incident dialysis patient populations pre- and post-eGFR reporting
Figure 1: Unadjusted interrupted time series of (A) mean eGFR, (B) proportion of patients initiated with an eGFR≥10.5 ml/min per m2, and (C) proportion of patients initiated dialysis with an eGFR≥12 ml/min per m2 before and after eGFR reporting. Each point represents the mean or proportion among incident dialysis patients in that 3-month period. Gray bars represent the 6-month transition periods. The solid line represents the linear trend line. Significance testing for the change in slope and the change in intercept: mean eGFR: P(slope)=0.66, P(intercept)=0.66; eGFR≥10.5 ml/min per m2: P(slope)=0.18, P(intercept)=0.47; eGFR≥12 ml/mi per m2: P(slope)=0.47, P(intercept)=0.57.
Table 3: Unadjusted change in the slope and intercept of the eGFR level over time among incident dialysis patient populations initiating dialysis before and after automated eGFR reporting
Association of eGFR Reporting and eGFR at Dialysis Initiation
In the multilevel adjusted models, there was no change in the slope of eGFR level at dialysis initiation over time among incident patient populations between the pre- and postreporting intervals (change in eGFR at dialysis initiation between pre- and postreporting β=0.01 [P=0.64] per 90-day period) (Table 4). This finding was similar in sensitivity analyses examining trends within the higher strata of eGFR (eGFR≥10.5: odds ratio, 1.02; 95% confidence interval, 1.0 to 1.05; P=0.09; eGFR≥12: odds ratio, 1.01; 95% confidence interval, 0.98 to 1.03; P=0.61). There were no significant changes in slope before or after eGFR reporting in any of the age and sex strata (P values for each stratum were nonsignificant) (Table 4). When stratified by province, there were no significant changes after eGFR reporting in Ontario (β=0.0, P=0.75 per 90-day period) and British Columbia (β=−0.01, P=0.80 per 90-day period). In Alberta, there was a decrease in the slope after eGFR reporting (β=−0.12, P<0.001 per 90-day period).
Table 4: Adjusted change in slope of the eGFR level at dialysis initiation among incident dialysis patient populations post-eGFR reported in age and sex subgroups
Discussion
In this large, population-level study of incident patients starting dialysis over the last decade, we found that laboratory-based automated eGFR reporting was not associated with changes in the eGFR level at dialysis initiation. The proportion of patients initiating dialysis with an eGFR≥10.5 or ≥12 ml/min per m2 did not change after eGFR reporting. There was no association with eGFR reporting and the level of eGFR at dialysis initiation in specific high-risk patient populations, such as the elderly, or women. Based on our findings, concern that widespread, automated eGFR reporting may be leading to initiation of dialysis at higher levels of eGFR are unwarranted.
This study is the first clinical study to investigate the relationship between eGFR reporting and the timing of dialysis initiation. There are plausible concerns for why eGFR reporting may be contributing to early dialysis initiation. The decision to start dialysis is not purely objective; rather, it depends on a combination of objective laboratory and clinical examination findings as well as a patient’s self-reported symptoms of uremia. Recognizing this subjectivity, clinical guidelines are deliberately vague (because of limited evidence) and merely suggest that dialysis be considered when eGFR drops below 15 ml/min per m2.19 Given this context, patient and physician biases may influence the decision to start dialysis. A portion of physicians has traditionally believed that starting dialysis earlier is better. For example, in a recent European survey of nephrologists, 11% felt that starting dialysis in patients with eGFR≥10.5 ml/min per m2 was always beneficial, even in the absence of symptoms.20 The clinical uncertainty about when to start dialysis was reported in Canada, where facility-level variation in eGFR at dialysis initiation was found to be 4.5% after accounting for case mix and facility-level factors.21 A recent international survey study randomized clinical vignettes presented to nephrologists to contain one of eGFR or serum creatinine and assessed opinions regarding the initiation of dialysis. The survey found that participants randomized to scenarios reporting the eGFR modestly increased the likelihood of initiating dialysis.22 There was no difference in responses from participants among the different countries (Canada included), and therefore, it is plausible that automated eGFR reporting might sway physicians to recommend dialysis earlier and more strongly, irrespective of symptom severity. Reasons for discrepant findings between the two studies may include limited generalizability in the survey study (response rate=19%), a relatively small number of respondents from Canada (n=119), and actual practice patterns differing from theoretical scenarios.
eGFR reporting has led to an increase in the detection of kidney disease, specifically in the elderly and women, and they are the same cohort susceptible to dialysis initiation at higher eGFR levels. Because elderly individuals may have a higher burden of comorbid conditions causing clinical symptoms, these individuals may be started early on dialysis, because nephrologists may falsely attribute symptoms to uremia rather than a coexisting condition. Reports have shown an increase of 3–6 years in the mean age of the referral population to nephrologists post-eGFR reporting.1,23,24 Concurrently, there have been increases of 6%–25% in women detected and referred to nephrology care with eGFR reporting.2,23 With advanced CKD, ongoing sarcopenia, malnutrition, and higher comorbidity have led to this same population being initiated on dialysis at higher levels of eGFR. In a United States dialysis cohort study of greater than 800,000 patients, the mean age of patients starting dialysis with an eGFR>15 ml/min per m2 was 68.4 versus 59.3 years in those patients started with an eGFR<5 ml/min per m2.25 A similar upward trend of age and initiating eGFR was reported in greater than 11,000 patients from the French Ramipril Efficacy in Nephropathy registry, where the mean age of those patients with an eGFR≤5 ml/min per m2 was 60.6 compared with 71.8 in those patients with an eGFR>20 ml/min per m2.14 Our findings may help in alleviating some of these concerns. Although the mean eGFR levels were significantly higher in the post-eGFR period (Table 2), there was no increase in the eGFR at dialysis initiation after eGFR reporting in the interrupted time series or the adjusted models. This finding was consistent when eGFR was examined as a continuous variable or proportions≥10.5 or ≥12 ml/min per m2 and in all subgroups (P value=NS), especially among those patients considered to be the most susceptible to early initiation (for example, elderly women). In one province (Alberta), it seemed that there was a decline in the eGFR level after eGFR reporting; however, reasons for this finding are unclear.
Initiation of dialysis at higher eGFR levels remains highly controversial. Numerous studies and a recent meta-analysis have consistently shown the lack of benefit and potential harm with early dialysis initiation.10,15,26–28 The Initiating Dialysis Early and Late (IDEAL) randomized control trial showed no difference in mortality between the two study arms: a high (10–14 ml/min per m2) and low eGFR (5–7 ml/min per m2) at dialysis initiation; however, there was a high degree of crossover between the two study groups.29 Furthermore, patients in the IDEAL trial started in the higher eGFR arm were initiated, on average, 6 months earlier than patients in the low eGFR arm, with a direct dialysis cost increase of greater than $10,000 per patient.30 eGFR reporting also has led to increases in resource use by increasing the referral rate from 3% to 270%.9 Subsequently, it would be concerning if eGFR reporting was associated with an increase in early dialysis initiation, because the benefit of early CKD detection, cardiovascular disease protection, and planning for dialysis would be offset by potential harm for those patients who initiate early dialysis and a significant increase in costs and health resource use.
Our findings strengthen the evidence for automated eGFR reporting as an important public health tool. To date, there is limited evidence on important health outcomes associated with eGFR reporting; evidence is often limited to short-term effects in CKD care, such as transient increases in the nephrologist referral rate, modest increases in angiotensin-converting enzyme inhibitor/angiotensin receptor blocker prescription, and a decline in nonsteroidal anti-inflammatory drug use.9,31,32 Our study is the first study to assess the association of eGFR reporting with a downstream, dialysis-related outcome, such as eGFR level at dialysis initiation. Additional studies to assess the association of eGFR reporting and dialysis-related outcomes, such as early optimization and adoption of home modalities and AVF creation, are required.
A major strength of our study is the unique implementation of eGFR reporting in Canada. Canada’s provincially administered health care system allowed for a staggered approach to eGFR reporting, thereby creating a natural experiment that enabled us to study the effects of reporting independent of era. Canada’s universal health care system is ideal for these types of population health studies in that all residents have access to a primary care system and in most cases, multidisciplinary kidney health prevention clinics. In other jurisdictions/countries, where eGFR reporting would be universally applied or applied at the level of each individual laboratory provider, such a study would not have been possible. Our findings are generalizable, because there was little missing data for eGFR at dialysis initiation (<7%); also, the findings represented population-level data of incident dialysis patients from the study provinces. Methodologically, we used an interrupted time series, a well established method of determining the temporal effects of an intervention. Furthermore, we formally examined the change in slope with multilevel models to avoid underestimation in the post-eGFR reporting period, a known limitation of using interrupted time series with upward-trending data.33,34 The 5-year follow-up window would adequately allow for patients to be detected by eGFR reporting, be referred, and progress to ESRD. A study conducted in two of four provinces included in our cohort (British Columbia and Ontario) reported that 24% of CKD patients reached kidney failure by approximately 3 years.35 Lastly, we were able to account for a substantive number of patient- and facility-level characteristics in our multivariable models because of the validated national registry Canadian Organ Replacement Registry (CORR), which tracks all new dialysis starts.
There are limitations to our work. There were possible unaccounted era effects (residual confounding); however, they would be minimized in our study, because the four provinces included initiated widespread eGFR reporting over an extended time period of 5 years (2003–2008), and we adjusted our model for calendar year. Laboratory creatinine measurements may not have been uniformly standardized with the implementation of eGFR reporting in each province. We attempted to minimize this lack of uniformity by the use of multilevel modeling that adjusted for facilities, thereby minimizing differences in the laboratory standardization of creatinine measurements.36 We lacked additional patient-level information, including the indication for dialysis initiation and patient functional measures.
In conclusion, widespread laboratory automated eGFR reporting was not associated with an increase in the eGFR level at dialysis initiation over time among incident dialysis patient populations. These findings help strengthen the evidence for eGFR reporting as an important public health initiative and alleviate concerns that it may be causing harm by leading to dialysis initiation at a higher level of eGFR, independent of uremic symptoms. Additional studies to establish the downstream impacts of eGFR reporting on dialysis-related outcomes are required.
Concise Methods
Population and Data Sources
All adult (>18 years old) incident dialysis patients from four Canadian provinces with automated laboratory-based eGFR reporting (Ontario, British Columbia, Alberta, and Newfoundland) with a measure of eGFR at dialysis initiation from January of 2001 to December 31, 2010, were included in the cohort (n=22,208; 93.4% of all incident patients started during that period). Data were obtained from the CORR, a validated registry that includes demographics, comorbidities, dialysis modality, vascular access, transplantation, and mortality on all ESRD patients in Canada.37,38 Data are collected by dialysis facilities and housed centrally with the Canadian Institute for Health Information. Data were collected until death, loss of follow-up, or end of study period (2010).
Cohort Definitions
Information regarding the timing and extent of province-wide automated eGFR reporting was obtained by surveying dialysis program heads and administrators. Provinces included in the study met the following criteria: (1) province-wide conversion to automated eGFR reporting, (2) conversion occurred within 1 year, and (3) there were adequate pre- and postintervention study periods (>1 year). The four provinces with automated eGFR reporting and the month and year of conversion were Ontario (January of 2006), Alberta (October of 2004), British Columbia (October of 2003), and Newfoundland (January of 2008). Manitoba and Nova Scotia also had eGFR reporting but were excluded because of a less than 1-year follow-up period (Manitoba implemented eGFR reporting in January of 2010) and a conversion period of greater than 1 year (occurred over 6 years in Nova Scotia). Because the four provinces had varying implementation time for eGFR reporting, we examined all patients initiating dialysis up to 5 years pre- and post-eGFR reporting. To allow for referrals and CKD care because of automated eGFR reporting, we excluded data from the first 6 months in the postreporting period (transition period).
First visit date with a nephrologist was used to estimate the length of predialysis nephrology care. Comorbidities (angina, chronic obstructive pulmonary disease, diabetes, malignancy, serious illness, hypertension, lung disease, coronary artery bypass grafting, pulmonary edema, peripheral vascular disease, stroke, cigarette smoking, and acute coronary syndrome) were recorded at the initiation of dialysis. Serious illness was defined as any illness that could shorten life expectancy to less than 5 years. Race was self-reported. Distance to facility was calculated as the direct linear distance in kilometers between a patient's primary residence (estimated from the postal code at the time of dialysis initiation) to the nearest dialysis provider using the formula by Vincenty.39 Facility variables included the proportion of patients who received dialysis through an AVF, whether the facility offered renal transplantation and/or peritoneal dialysis, and the center size. Individual patients and dialysis facilities were deidentified for analytic purposes. Additional laboratory values (hemoglobin, albumin, and phosphate) were recorded in cross-section at dialysis initiation.
Outcome
The primary outcome was the change in eGFR level at dialysis initiation over time among incident dialysis patient populations. Kidney function was estimated using the abbreviated Modification of Diet in Renal Disease equation based on the patients serum creatinine, age, and sex at dialysis initiation.40 The secondary outcome was the eGFR level over time among patients at dialysis initiation stratified by age in tertiles and sex.
Statistical Analyses
Patient and facility characteristics were compared between the pre- and post-eGFR reporting periods. Continuous variables of interest were summarized as mean or median with SD or interquartile range as appropriate. Differences in characteristics were determined by t test or the Mann–Whitney test for continuous variables and chi-squared or the Fisher-exact test for dichotomous variables.
The association between eGFR reporting and the eGFR level at dialysis initiation over time among incident dialysis patient populations was examined by two methods. First, an unadjusted interrupted time series was created for the time interval of 5 years before and after eGFR reporting.33 Time was divided into 3-month intervals, and the quarterly average or proportion was examined for trends. Second, sequential adjusted multilevel regression models were used to examine changes in the eGFR at dialysis initiation in the pre- and postreporting time intervals.41,42 Changes in trends among the populations were measured by examining changes in the slope of the eGFR at dialysis initiation after eGFR reporting using an interaction term (model equation is presented in Supplemental Appendix). Two levels were included in the multilevel model: patients and facilities. A two-level model was used to account for significant facility-level variability in the eGFR at dialysis initiation, a finding that we had previously reported.21 Fully adjusted models included variables for case mix (age, sex, body mass index, race, comorbidities, laboratory values, distance from center, and modality), facility-level variables (percent AVF, transplant facility, peritoneal dialysis facility, and center size), and calendar year. In a series of sensitivity analyses, we also examined the primary outcome as the proportion of patients with an eGFR at dialysis initiation≥10.5 and ≥12 ml/min per m2. Results were further stratified by age in tertiles, sex, and province. When examining the provincial subgroups, only Ontario, Alberta, and British Columbia were examined, because Newfoundland had a small number of facilities.
Multiple imputation was used for missing values.43 Analyses were performed using PASW version 18 and SAS version 9.3. All hypothesis tests were two-sided, with statistical significance determined at a P value <0.05.
Disclosures
M.M.S. has been on advisory boards for Roche and Amgen and received speaking honoraria from Roche, Amgen, and Sanofi US.
We thank Bob Williams and Frank Ivis at the Canadian Organ Replacement Registry.
M.M.S. has provided salary support by the Jindal Research Chair for the Prevention of Kidney Disease at the University of Ottawa.
This work was presented as an oral abstract at the American Society of Nephrology Annual General Meeting in Atlanta, GA, on November 9, 2013.
M.M.S. had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Published online ahead of print. Publication date available at www.jasn.org.
This article contains supplemental material online at http://jasn.asnjournals.org/lookup/suppl/doi:10.1681/ASN.2013090953/-/DCSupplemental.
References
1. Noble E, Johnson DW, Gray N, Hollett P, Hawley CM, Campbell SB, Mudge DW, Isbel NM: The impact of automated eGFR reporting and education on nephrology service referrals. Nephrol Dial Transplant 23: 3845–3850, 2008
2. Jain AK, McLeod I, Huo C, Cuerden MS, Akbari A, Tonelli M, van Walraven C, Quinn RR, Hemmelgarn B, Oliver MJ, Li P, Garg AX: When laboratories report estimated glomerular filtration rates in addition to serum creatinines, nephrology consults increase. Kidney Int 76: 318–323, 2009
3. Accetta NA, Gladstone EH, DiSogra C, Wright EC, Briggs M, Narva AS: Prevalence of estimated GFR reporting among US clinical laboratories. Am J Kidney Dis 52: 778–787, 2008
4. Richards N, Harris K, Whitfield M, O'Donoghue D, Lewis R, Mansell M, Thomas S, Townend J, Eames M, Marcelli D: The impact of population-based identification of chronic kidney disease using estimated glomerular filtration rate (eGFR) reporting. Nephrol Dial Transplant 23: 556–561, 2008
5. Akbari A, Grimshaw J, Stacey D, Hogg W, Ramsay T, Cheng-Fitzpatrick M, Magner P, Bell R, Karpinski J: Change in appropriate referrals to nephrologists after the introduction of automatic reporting of the estimated glomerular filtration rate. CMAJ 184: E269–E276, 2012
6. Glassock RJ, Winearls C: Screening for CKD with eGFR: Doubts and dangers. Clin J Am Soc Nephrol 3: 1563–1568, 2008
7. Clase CM, Garg AX, Kiberd BA: Classifying kidney problems: Can we avoid framing risks as diseases? BMJ 392: 912–915, 2004
8. Hemmelgarn BR, Zhang J, Manns BJ, James MT, Quinn RR, Ravani P, Klarenbach SW, Culleton BF, Krause R, Thorlacius L, Jain AK, Tonelli MAlberta Kidney Disease Network: Nephrology visits and health care resource use before and after reporting estimated glomerular filtration rate. JAMA 303: 1151–1158, 2010
9. Kagoma YK, Weir MA, Iansavichus AV, Hemmelgarn BR, Akbari A, Patel UD, Garg AX, Jain AK: Impact of estimated GFR reporting on patients, clinicians, and health-care systems: A systematic review. Am J Kidney Dis 57: 592–601, 2011
10. O’Hare AM, Choi AI, Boscardin WJ, Clinton WL, Zawadzki I, Hebert PL, Kurella Tamura M, Taylor L, Larson EB: Trends in timing of initiation of chronic dialysis in the United States. Arch Intern Med 171: 1663–1669, 2011
11. Clark WF, Na Y, Rosansky SJ, Sontrop JM, Macnab JJ, Glassock RJ, Eggers PW, Jackson K, Moist L: Association between estimated glomerular filtration rate at initiation of dialysis and mortality. CMAJ 183: 47–53, 2011
12. Stel VS, Tomson C, Ansell D, Casino FG, Collart F, Finne P, Ioannidis GA, De Meester J, Salomone M, Traynor JP, Zurriaga O, Jager KJ: Level of renal function in patients starting dialysis: An ERA-EDTA Registry study. Nephrol Dial Transplant 25: 3315–3325, 2010
13. Stel VS, Dekker FW, Ansell D, Augustijn H, Casino FG, Collart F, Finne P, Ioannidis GA, Salomone M, Traynor JP, Zurriaga O, Verrina E, Jager KJ: Residual renal function at the start of dialysis and clinical outcomes. Nephrol Dial Transplant 24: 3175–3182, 2009
14. Lassalle M, Labeeuw M, Frimat L, Villar E, Joyeux V, Couchoud C, Stengel B: Age and comorbidity may explain the paradoxical association of an early dialysis start with poor survival. Kidney Int 77: 700–707, 2010
15. Rosansky SJ, Eggers P, Jackson K, Glassock R, Clark WF: Early start of hemodialysis may be harmful. Arch Intern Med 171: 396–403, 2011
16. Kazmi WH, Gilbertson DT, Obrador GT, Guo H, Pereira BJ, Collins AJ, Kausz AT: Effect of comorbidity on the increased mortality associated with early initiation of dialysis. Am J Kidney Dis 46: 887–896, 2005
17. Hwang SJ, Yang WC, Lin MY, Mau LW, Chen HCTaiwan Society of Nephrology: Impact of the clinical conditions at dialysis initiation on mortality in incident haemodialysis patients: A national cohort study in Taiwan. Nephrol Dial Transplant 25: 2616–1624, 2010
18. Bao YDL, Dalrymple L, Chertow GM, Kaysen GA, Johansen KL: Frailty, dialysis initiation, and mortality in end-stage renal disease. Arch Intern Med 172: 1071–1077, 2012
19. National Kidney Foundation: KDOQI clinical practice guidelines and clinical practice recommendations for 2006 updates: Hemodialysis adequacy, peritoneal dialysis adequacy and vascular access. Am J Kidney Dis 48[Suppl 1]: S1–S322, 2006
20. van de Luijtgaarden MWM, Noordzij M, Tomson C, Couchoud C, Cancarini G, Ansell D, Bos WJ, Dekker FW, Gorriz JL, Iatrou C, Garneata L, Wanner C, Cala S, Stojceva-Taneva O, Finne P, Stel VS, van Biesen W, Jager KJ: Factors influencing the decision to start renal replacement therapy: Results of a survey among European nephrologists. Am J Kidney Dis 60: 940–948, 2012
21. Sood MM, Manns B, Dart A, Hiebert B, Kappel J, Komenda P, Molzahn A, Naimark D, Nessim S, Rigatto C, Soroka S, Zappitelli M, Tangri N: Geographic and facility-level variation in the level of eGFR at dialysis initiation. Presented at the 50th ERA-EDTA Congress, Istanbul, Turkey, May 18, 2013
22. Brimble KS, Mehrotra R, Tonelli M, Hawley CM, Castledine C, McDonald SP, Levidiotis V, Gangji AS, Treleaven DJ, Margetts PJ, Walsh M: Estimated GFR reporting influences recommendations for dialysis initiation. J Am Soc Nephrol 24: 1737–1742, 2013
23. Shields JMR, Morgan N, Ryan M, Harron JC: Effect of introduction of eGFR reporting to primary care on referral rates and quality and appropriateness of referrals to specialist nephrology services [abstract]. Presented at the American Society of Nephrology Conference, Philadelphia, November 5–9, 2006
24. Phillips LA, Donovan KL, Phillips AO: Renal quality outcomes framework and eGFR: Impact on secondary care. QJM 102: 415–423, 2009
25. Wright S, Klausner D, Baird B, Williams ME, Steinman T, Tang H, Ragasa R, Goldfarb-Rumyantzev AS: Timing of dialysis initiation and survival in ESRD. Clin J Am Soc Nephrol 5: 1828–1835, 2010
26. Sawhney S, Djurdjev O, Simpson K, Macleod A, Levin A: Survival and dialysis initiation: Comparing British Columbia and Scotland registries. Nephrol Dial Transplant 24: 3186–3192, 2009
27. Susantitaphong P, Altamimi S, Ashkar M, Balk EM, Stel VS, Wright S, Jaber BL: GFR at initiation of dialysis and mortality in CKD: A meta-analysis. Am J Kidney Dis 59: 829–840, 2012
28. Beddhu S, Samore MH, Roberts MS, Stoddard GJ, Ramkumar N, Pappas LM, Cheung AK: Impact of timing of initiation of dialysis on mortality. J Am Soc Nephrol 14: 2305–2312, 2003
29. Cooper BA, Branley P, Bulfone L, Collins JF, Craig JC, Fraenkel MB, Harris A, Johnson DW, Kesselhut J, Li JJ, Luxton G, Pilmore A, Tiller DJ, Harris DC, Pollock CAIDEAL Study: A randomized, controlled trial of early versus late initiation of dialysis. N Engl J Med 363: 609–619, 2010
30. Harris A, Cooper BA, Li JJ, Bulfone L, Branley P, Collins JF, Craig JC, Fraenkel MB, Johnson DW, Kesselhut J, Luxton G, Pilmore A, Rosevear M, Tiller DJ, Pollock CA, Harris DC: Cost-effectiveness of initiating dialysis early: A randomized controlled trial. Am J Kidney Dis 57: 707–715, 2011
31. Kagoma YK, Weir MA, Iansavichus AV, Hemmelgarn BR, Akbari A, Patel UD, Garg AX, Jain AK: Impact of estimated GFR reporting on patients, clinicians, and health-care systems: A systematic review. Am J Kidney Dis 57: 592–601, 2011
32. Wei L, MacDonald TM, Jennings C, Sheng X, Flynn RW, Murphy MJ: Estimated GFR reporting is associated with decreased nonsteroidal anti-inflammatory drug prescribing and increased renal function. Kidney Int 84: 174–178, 2013
33. Ramsay C, Brown E, Hartman G, Davey P: Room for improvement: A systematic review of the quality of evaluations of interventions to improve hospital antibiotic prescribing. J Antimicrob Chemother 52: 764–771, 2003
34. Wagner AK, Soumerai SB, Zhang F, Ross-Degnan D: Segmented regression analysis of interrupted time series studies in medication use research. J Clin Pharm Ther 27: 299–309, 2002
35. Tangri N, Stevens LA, Griffith J, Tighiouart H, Djurdjev O, Naimark D, Levin A, Levey AS: A predictive model for progression of chronic kidney disease to kidney failure. JAMA 305: 1553–1559, 2011
36. Komenda P, Beaulieu M, Seccombe D, Levin A: Regional implementation of creatinine measurement standardization. J Am Soc Nephrol 19: 164–169, 2008
37. Canadian Institute for Health Information: Treatment of end-stage organ failure in Canada, 1999 to 2008—CORR 2010 Annual Report. Ottawa, ON, Canada, 2010
38. Moist LM, Richards HA, Miskulin D, Lok CE, Yeates K, Garg AX, Trpeski L, Chapman A, Amuah J, Hemmelgarn BR: A validation study of the Canadian Organ Replacement Register. Clin J Am Soc Nephrol 6: 813–818, 2011
39. Vincenty T: Direct and inverse solutions of geodesics on the ellipsoid with application of nested equations. Survey Rev 23: 88–93, 1975
40. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth DModification of Diet in Renal Disease Study Group: A more accurate method to estimate glomerular filtration rate from serum creatinine: A new prediction equation. Ann Intern Med 130: 461–470, 1999
41. Greenland S: Principles of multilevel modelling. Int J Epidemiol 29: 158–167, 2000
42. Rodríguez G, Elo I: Intra-class correlation in random-effects models for binary data. Stata J 3: 32–46, 2003
43. Schafer JL: Multiple imputation: A primer. Stat Methods Med Res 8: 3–15, 1999
