Significance Statement
In 2009, the Initiating Dialysis Early and Late (IDEAL) trial found no clinically measurable benefit with early dialysis initiation, but whether these findings were widely adopted was unknown. The authors conducted a cluster randomized trial, with 55 clinics randomized to the intervention (a multifaceted knowledge translation intervention aimed at promoting an intent-to-defer strategy for dialysis initiation) and control. In their analysis, which included 3424 patients initiating dialysis in the 1-year follow-up period, they found no statistically significant difference between the two groups in the proportion of patients who initiated dialysis early (at eGFR>10.5ml/min per 1.73m2) or in the proportion of patients who initiated dialysis as an acute inpatient. The knowledge translation intervention failed to further reduce the proportion of early dialysis starts in multidisciplinary CKD clinics.
Keywords: randomized controlled trials, clinical trial, chronic dialysis, glomerular filtration rate, hospitalization, kidney failure
Abstract
Background
The Initiating Dialysis Early and Late (IDEAL) trial, published in 2009, found no clinically measurable benefit with respect to risk of mortality or early complications with early dialysis initiation versus deferred dialysis start. After these findings, guidelines recommended an intent-to-defer approach to dialysis initiation, with the goal of deferring it until clinical symptoms arise.
Methods
To evaluate a four-component knowledge translation intervention aimed at promoting an intent-to-defer strategy for dialysis initiation, we conducted a cluster randomized trial in Canada between October 2014 and November 2015. We randomized 55 clinics, 27 to the intervention group and 28 to the control group. The educational intervention, using knowledge-translation tools, included telephone surveys from a knowledge-translation broker, a 1-year center-specific audit with feedback, delivery of a guidelines package, and an academic detailing visit. Participants included adults who had at least 3 months of predialysis care and who started dialysis in the first year after the intervention. The primary efficacy outcome was the proportion of patients who initiated dialysis early (at eGFR >10.5 ml/min per 1.73 m2). The secondary outcome was the proportion of patients who initiated in the acute inpatient setting.
Results
The analysis included 3424 patients initiating dialysis in the 1-year follow-up period. Of these, 509 of 1592 (32.0%) in the intervention arm and 605 of 1832 (33.0%) in the control arm started dialysis early. There was no difference in the proportion of individuals initiating dialysis early or in the proportion of individuals initiating dialysis as an acute inpatient.
Conclusions
A multifaceted knowledge translation intervention failed to reduce the proportion of early dialysis starts in patients with CKD followed in multidisciplinary clinics.
Clinical Trial registry name and registration number:
ClinicalTrials.gov, NCT02183987. Available at: https://clinicaltrials.gov/ct2/show/NCT02183987
In the past, there was uncertainty over the optimal level of kidney function at which to initiate dialysis therapies to improve clinical outcomes. This led to significant variation in the timing of dialysis initiation, and in Canada, the proportion of all patients on dialysis who initiated dialysis early (defined as starting dialysis with an eGFR over 10.5 ml/min per 1.73 m2) steadily increased from 28% to 36% between 2001 and 2007.1 These secular changes in the timing of dialysis initiation led to a greater number of incident and prevalent patients receiving dialysis, with greater overall treatment burden and health care costs.2 Similar trends were observed in the United States, with 19% initiating dialysis early in 1996, increasing to 45% by 2005.3
In 2009, Cooper et al. published the Initiating Dialysis Early and Late (IDEAL) trial and found no clinically measurable benefit with early dialysis initiation. Specifically, the investigators found no differences in the risk of mortality or early complications with early versus deferred dialysis start, and higher costs were associated with an earlier start.4 As such, clinical practice guidelines in Canada and worldwide recommended an intent-to-defer approach to dialysis initiation, with the goal of deferring initiation until clinical symptoms arise.5
However, barriers to an intent-to-defer approach are common, and include physician beliefs about the health benefits of early initiation, a more fragile and elderly dialysis population with sarcopenia confounding eGFR measurement, the wish to initiate dialysis in a controlled rather than emergency situation, and a preference for early initiation in patients considering home dialysis.6,7 Strategies to overcome these barriers and ensure evidence gets into practice are required because passive dissemination of clinical practice guidelines in other areas of medicine leads to most guideline recommendations being ignored.8
Here, we present findings from a national, multicenter, two-arm parallel design, open-label, cluster-randomized trial of a multifaceted knowledge-translation intervention designed to promote an intent-to-defer strategy for dialysis initiation compared with routine care. Our primary efficacy outcome was the between-group difference of proportion of patients initiating dialysis with an eGFR >10.5 ml/min per 1.73 m2 (a proxy for early dialysis initiation). The primary safety outcome included the proportion of patients who started dialysis acutely in hospital.
Methods
Study Design
A detailed study protocol has previously been published.9 In brief, we compared the effectiveness of the multifaceted knowledge-translation strategy (intervention group) versus simple guideline release without active dissemination (control group) on the timing of dialysis initiation in patients from 55 advanced CKD clinics (clusters) across nine provinces in Canada. In most Canadian provinces, care in these clinics is directed by nephrologists in concert with an interdisciplinary team (nurse, dietician, pharmacist, and/or social work support), and include patients with advanced CKD (eGFR <30 ml/min per 1.73 m2) or an estimated risk of over 10% of receiving dialysis or a kidney transplant within 2 years, on the basis of the Kidney Failure Risk Equation.10
Randomization was performed at the level of the multidisciplinary CKD clinic (versus at the level of a nephrologist or patient) to avoid experimental contamination (e.g., nephrologists who care for patients with advanced CKD located at the same clinic), with stratification variables including the region and size of the CKD clinic (<200 patients, 200–600 patients, and >600 patients seen in the prior year). A simulation of randomization was completed to ensure balance on cluster-level variables, with a randomization that resulted in good balance on all measured cluster factors being selected. After randomization, the intervention (see below for details) was applied in a staggered roll-out between October 2014 and November 2015, with time zero for the intervention clinics defined as the date of the medical detailing visit for the intervention clinic, and time zero for the control clinics defined as the mean date of the detailing visits in the same randomized block. The trial received approval from the Health Research Ethics Board at the University of Manitoba and the Conjoint Health Research Ethics Board at the University of Calgary. Because the educational intervention was aimed at providers and promoting the use of an accepted clinical guideline, the research ethics boards felt the risk to patients was low and that individual patient consent was not necessary.
Data Sources and Participants
Data on dialysis starts were collected from the Canadian Organ Replacement Register (CORR). CORR is a government-mandated, validated organ failure registry that tracks all information on organ replacement therapy in Canada, including dialysis activity from initiation until death. Patients from the province of Quebec were excluded as privacy laws preclude submission of deidentified data without first-person consent, and as such the province’s data were delayed or incomplete.11 Data extracted from CORR included patient demographics (age, sex), clinical characteristics (body mass index, comorbidities, etiology of kidney failure), laboratory measurements at initiation of dialysis (serum creatinine and eGFR, serum hemoglobin, serum phosphate, and serum albumin), information on predialysis care (date of dialysis start and duration of treatment), and location of treatment. The setting of first dialysis treatment (acute start in hospital versus outpatient) was obtained by linking CORR to the Canadian Institute for Health Information’s Discharge Abstract Database12 to ascertain hospitalizations, and was available for all of Canada except the province of Manitoba, due to differences in health card number encryption that did not allow for data linkage. Data collection is performed as part of routine reporting requirements to the Canadian Institute of Health Information by personnel unaware of the trial conduct, and data analysts were blinded to the assignment of the intervention.
For the primary outcome, we included all patients who initiated chronic dialysis in the first year after delivery of the intervention and had at least 3 months of predialysis care. Patients aged <18 years, those who received a pre-emptive kidney transplant, those who recovered kidney function within the first 90 days of dialysis initiation (suggestive of AKI), and those who had a recorded eGFR >30 ml/min per 1.73 m2 at dialysis initiation were excluded (indicative of a data entry error).
Intervention
We evaluated a four-component educational intervention using active knowledge-translation tools, each included to address a specific barrier identified in a cross-Canada survey of nephrologists and multidisciplinary clinics eliciting approaches to dialysis initiation.13 We used the services of a knowledge-translation broker with over a decade of experience in knowledge translation, knowledge management, and outreach at both the provincial and national level. First, the knowledge-translation broker initiated contact with the clinic and introduced the initiative of preventing early dialysis starts and conducted a telephone survey to gather information on current clinic practice and flow. Second, we provided a 1-year center-specific audit and feedback report detailing the proportion of patients initiating dialysis with an eGFR >10.5 ml/min per 1.73 m2 using data from CORR. Next, clinics were sent information packages including Canadian Society of Nephrology practice guidelines on the optimal timing of dialysis initiation,5 a provider-directed infographic, a patient-directed infographic (which clinics were asked to post in clinics), patient handouts (which clinics were asked to distribute to patients approaching the need for dialysis), and a specially developed educational video. Clinics were advised to play the video in patient waiting areas to be easily accessible. After this, a one-time site visit was conducted by a visiting nephrologist who provided educational rounds on evidence and guidance supporting the intent-to-defer strategy, including a review of the clinic’s current dialysis initiation practice (in comparison with other provincial clinics as audit and feedback), explanation on using provided tools, and identification of a local champion at each clinic was the aim of this visit. This provider-facing education was provided to all of the treating nephrologists at the clinic and the clinic staff. The educational material was developed by consulting a panel of patients, nephrologists, and knowledge-translation experts in 2013, with the initial four academic detailing visits in the study provided by the co-Principal Investigators (Drs. Tangri and Manns), with a refining of educational materials on the basis of feedback from these visits. Additional details are available in the study protocol.9 The audit and feedback design has been demonstrated to be an effective tool in reducing practice pattern variation in the management of chronic conditions.14,15
After the implementation of the intervention, follow-up was conducted on the clinics included in the intervention arm. The knowledge-translation broker was in regular contact with the clinics by email and telephone over the 12-month period after the clinics’ medical detailing visit. Semistructured interviews and surveys were conducted by the knowledge-translation broker to ascertain use and experience with the tools, experience with the medical detailing visit, and feedback about future guidelines, types of support required, and any other outstanding issues. In addition, the level of clinic engagement for adoption of the guidelines was assessed via Likert scale by the knowledge-translation broker. Lastly, the visiting nephrologist followed up with the clinic to reinforce key messages, information on how tools were applied in clinics, and to address any additional questions or concerns.
A diagram outlining the components of the intervention is presented in Figure 1.
Figure 1.
Overview of knowledge translation, audit and feedback, medical detailing, and ongoing follow-up. KT, knowledge translation.
Control
Clinics in the control group received none of the four educational active knowledge translation interventions outlined above. They may have had access to publicly available information such as journal articles, conference presentations, or the timing of dialysis initiation guideline;5 however, this information was not actively disseminated to clinics in this group.
Outcomes
Outcomes were obtained at the patient level and clustered by predialysis clinic group through administrative databases. The primary efficacy outcome of the study was the proportion of patients on dialysis who initiated it early (eGFR >10.5 ml/min per 1.73 m2) during the 1-year follow-up period with eGFR calculated using the Modification of Diet in Renal Disease study equation.16 The primary safety outcome was the proportion of incident patients receiving dialysis who initiated in the acute inpatient setting, termed “crash starts.”
Statistical Analyses
The dichotomous outcomes in our study were analyzed using the two-sample t test, using generalized estimated equations (GEEs) to obtain the population average affect in the presence of clustering by applying a multilevel model accounting for both individual- and cluster-level effects. The log-binomial model was used when possible, and for nonconvergent models we applied a log-Poisson model. Similarly, for continuous outcomes we applied a GEE model with a normal distribution and identity link.17 We also conducted a stratified analysis to align with the stratified randomization in our design, but because the results were consistent with the nonstratified models, we presented the more parsimonious approach.
Sensitivity Analyses
We conducted several additional analyses. We examined the differences-in-differences approach to control for differences between the intervention and control group in the pretrial period and secular trends over time.18 As such, we constructed GEE models where the changes over time are compared using the interaction between group (intervention and control) and time (postintervention versus preintervention).We repeated the primary analysis adjusted for patient-level characteristics (age, sex, comorbidities, ethnicity, concurrent laboratory values at dialysis initiation, and distance from the patient’s residence to the dialysis treatment center) with missing covariate data imputed over five iterations using SAS software’s PROC MI. Any covariates having more than 30% of values missing were not imputed and excluded from the analysis. Lastly, a post-hoc analysis included only clinics that were found in the follow-up survey to have been high users of the knowledge-translation tools, defined as those who responded to a 6-month follow-up telephone-based survey were rated by the knowledge translation broker as adhering to the intervention with a score ≥ three out of five on a Likert scale.
Power Calculation
We had over 90% power to detect an absolute risk reduction in early dialysis starts of 10%, assuming a baseline of 40% of patients initiating early in the control group, an intraclass correlation coefficient of 0.031, 55 clusters, an estimated 3696 dialysis starts in the year post-trial, and an alpha of 0.05.9
All analyses were performed using SAS software, Version 9.4 (Cary, NC). A two-sided P value of 0.05 was used to determine statistical significance.
Results
Study Population
All Canadian CKD clinics (excluding those in Quebec) were randomized, totaling 55 clinics. Of these clinics, 27 were randomized to the intervention and 28 to the control. The intervention was delivered as intended in 26 out of 27 clinics, with only one clinic not receiving an academic detailing visit. All intervention clinics received telephone calls or email communication from the study team highlighting the goals of the study, and the intent of the audit and feedback reports.
Among the 27 clinics randomized to the intervention arm, 2476 patients started dialysis in the 1-year follow-up period. Of these patients, 373 were excluded for not having a recorded eGFR at dialysis initiation, 31 were excluded for having a recorded eGFR at dialysis initiation >30 ml/min per 1.73 m2, and 480 were excluded for having <3 months of predialysis nephrology care. Similarly, among the 28 clinics randomized to the control arm, 2836 patients initiated dialysis during the follow-up period, 263 were excluded for not having a recorded eGFR, 43 were excluded for having a recorded eGFR at dialysis initiation >30 ml/min per 1.73 m2, and 698 were excluded for not having 3 months of predialysis nephrology care (Figure 2).
Figure 2.
Overview of randomization, intervention assignment, exclusion criteria, and follow-up.
Characteristics of clusters were similar between the intervention and control groups, with a mean of 770±667 patients for clinics in the treatment arm and 699±598 patients in clinics in the control arm, and a mean of 6.4 nephrologists per clinic in both groups. For individuals that initiated dialysis during the study period, demographic characteristics and comorbid conditions were similar in the control and intervention clinics. Characteristics of the included patients and facilities are described in Table 1.
Table 1.
Baseline characteristics of clinics and patients
| Characteristics | Intervention | Control |
|---|---|---|
| Clinics | n=27 | n=28 |
| Number of patients (SD) | 770 (667) | 699 (598) |
| Number of nephrologists (SD) | 6.4 (5.0) | 6.4 (6.5) |
| Cluster size | ||
| <200 patients | 3 (11.1%) | 2 (7.1%) |
| 200–600 patients | 11 (40.7%) | 13 (46.4%) |
| >600 patients | 13 (48.1%) | 13 (46.4%) |
| Model of care | ||
| Rotating nurse and rotating nephrologist | 2 (7.4%) | 3 (10.7%) |
| Same nurse and rotating nephrologist | 3 (11.1%) | 2 (7.1%) |
| Rotating nurse and same nephrologist | 6 (22.2%) | 5 (17.9%) |
| Same nurse and same nephrologist | 15 (55.6%) | 17 (60.7%) |
| Mixed | 1 (3.7%) | 1 (3.6%) |
| Region | ||
| Alberta | 2 (7.4%) | 2 (7.1%) |
| British Columbia | 6 (10.9%) | 6 (21.4%) |
| Maritime provinces | 4 (7.3%) | 4 (14.3%) |
| Prairie provinces (Manitoba/Saskatchewan) | 3 (5.5%) | 3 (10.7%) |
| Ontario | 12 (21.8%) | 13 (46.4%) |
| Patients | n=1592 | n=1832 |
| Demographics and labs | ||
| Age (SD) | 63.3 (0.4) | 64.9 (0.3) |
| Men | 1012 (63.6%) | 1113 (60.8%) |
| BMI (kg/m2) | 29.2 (0.2) | 29.0 (0.2) |
| eGFR (ml/min per 1.73 m2) | 9.44 (0.10) | 9.71 (0.10) |
| Hemoglobin (g/L) | 97.2 (0.4) | 95.7 (0.4) |
| Phosphate (mmol/L) | 1.92 (0.01) | 1.86 (0.01) |
| Albumin (g/L) | 33.5 (0.2) | 32.1 (0.2) |
| Comorbidities | ||
| Angina | 206 (12.9%) | 256 (14.0%) |
| Previous CABG | 241 (15.1%) | 372 (20.3%) |
| Cerebrovascular accident | 221 (13.9%) | 247 (13.5%) |
| Current smoker | 223 (14.0%) | 237 (12.9%) |
| Diabetes mellitus | 1036 (65.1%) | 1162 (63.4%) |
| Hypertension | 1270 (79.8%) | 1503 (82.0%) |
| Lung disease | 175 (11.0%) | 248 (13.5%) |
| Malignant neoplasm | 227 (14.3%) | 301 (16.4%) |
| Myocardial infarction | 281 (17.7%) | 378 (20.6%) |
| Peripheral vascular disease | 237 (14.9%) | 317 (17.3%) |
| Pulmonary edema | 385 (24.2%) | 477 (26.0%) |
| Cause of ESKD | ||
| Hypertension | 160 (10.1%) | 241 (13.2%) |
| Diabetes mellitus | 844 (53.0%) | 888 (48.5%) |
| GN | 193 (12.1%) | 195 (10.6%) |
| Obstruction | 16 (1.0%) | 27 (1.5%) |
| Interstitial | 52 (3.3%) | 77 (4.2%) |
| Polycystic kidney disease | 103 (6.5%) | 99 (5.4%) |
| Other | 224 (14.1%) | 305 (16.7%) |
| Race/ethnicity | ||
| White | 1098 (69.0%) | 1329 (72.5%) |
| East Asian | 147 (9.2%) | 143 (7.8%) |
| Indigenous | 137 (8.6%) | 113 (6.2%) |
| South Asian | 107 (6.7%) | 163 (8.9%) |
| Black | 60 (3.8%) | 67 (3.7%) |
| Other | 43 (2.7%) | 17 (0.9%) |
| Distance from patient’s residence to center (km) | ||
| <50 | 1247 (78.3%) | 1506 (82.2%) |
| 50–100 | 244 (15.3%) | 205 (11.2%) |
| >150 | 101 (6.3%) | 121 (6.6%) |
SD, standard deviation; BMI, body mass index; CABG, coronary artery bypass graft.
Primary Outcome
In the year after the active knowledge-translation intervention, 509 (32.0%) patients in the intervention arm and 605 (33.0%) patients in the control arm started dialysis with an eGFR >10.5 ml/min per 1.73 m2. Our main analysis, accounting for the presence of clustering, found no difference in the proportion of patients initiating dialysis early in the postintervention period between the two groups (absolute difference, −2.6%; 95% confidence interval [95% CI], −11.7% to 6.5%; P=0.58). The relative risk between the two groups accounting for clustering was 0.93 (95% CI, 0.71 to 1.21; P=0.58) (Table 2). Mean eGFR at dialysis initiation was not statistically different between the two groups in the postintervention period, at 9.50±0.26 in the intervention arm and 9.94±0.31 in the control arm (P=0.27).
Table 2.
Results of primary analysis evaluating the outcomes of early dialysis initiation and acute inpatient initiation with GEE models in the yr after KT intervention
| Outcome | Intervention (n=1592) | Control (n=1832) | Relative Risk (95% CI) | Absolute Risk Difference (95% CI) | P value |
|---|---|---|---|---|---|
| Early dialysis initiation (eGFR >10.5 ml/min per 1.73 m2) | 509 (32.0%) | 605 (33.0%) | 0.93 (0.71 to 1.21) | −2.6% (−11.7% to 6.5%) | 0.58 |
| Initiated dialysis in hospital | 448/1515 (29.6%) | 519/1713 (30.3%) | 0.96 (0.83 to 1.11) | −1.2% (−5.6% to 3.3%) | 0.61 |
KT, knowledge translation.
Secondary Outcomes
The primary safety outcome was analyzed in 3228 patients in the post-trial period. We observed 448 (29.6%) of the patients in the intervention arm and 519 (30.3%) of the patients in the control arm starting dialysis as an acute inpatient in the year after the knowledge-translation intervention. After adjustment for the presence of clustering, there was no difference in the proportion initiating dialysis as an acute inpatient in the postintervention period between the two groups (absolute difference, −1.2%; 95% CI, −5.6% to 3.3%; P=0.61). The relative risk between the two groups accounting for clustering was 0.96 (95% CI, 0.83 to 1.11; P=0.61) (Table 2).
Sensitivity Analyses
We found similar findings to the primary analysis when applying a differences-in-differences approach, with no statistically significant intervention effect with respect to early dialysis initiations (relative risk, 1.10; 95% CI, 0.97 to 1.24; P=0.15) and no statistically significant intervention effect with respect to acute in-hospital dialysis initiations (relative risk, 1.08; 95% CI, 0.92 to 1.26; P=0.34) (Table 3). Additional information on descriptive statistics of patients in the pre- and post-trial periods for the differences-in-differences approach is provided in Supplemental Material 1, and an overview of randomization and follow-up is provided in Supplemental Material 2.
Table 3.
Results of differences-in-differences analyses with GEE models comparing the 1-yr pre-KT intervention to the 1-yr post-KT intervention
| Outcome | Intervention Group | Control Group | Effect of Intervention | |||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Preintervention (n=1791) | Postintervention (n=1592) | Difference | Preintervention (n=1833) | Postintervention (n=1832) | Difference | Crude Absolute Risk Difference between Groups | Absolute Risk Difference between Groups Accounting for Clustering | Relative Risk between Groups Accounting for Clustering | P value | |
| number (percent) | percent (95% CI) | number (percent) | Percent (95% CI) | percent (95% CI) | Percent (95% CI) | relative risk (95% CI) | ||||
| Early dialysis initiation (eGFR> 10.5 ml/min per 1.73 m2) | 533 (29.8) | 509 (32.0) | 2.2% (−0.9% to 5.3%) | 611 (33.3) | 605 (33.0) | −0.3% (−3.4% to 2.7%) | 2.5% (−1.8% to 6.9%) | 3.1% (−1.2% to 7.3%) | 1.10 (0.97 to 1.24) | 0.15 |
| Initiated dialysis in hospital | 487/1708 (28.5) | 448/1515 (29.6) | 1.1% (−2.1% to 4.2%) | 531/1692 (31.4) | 519/1713 (30.3) | −1.1% (−4.2% to 2.0%) | 2.1% (−2.3% to 6.6%) | 2.3% (−2.4% to 7.0%) | 1.08 (0.92 to 1.26) | 0.34 |
KT, xx.
In fully adjusted analyses (adjusted for age; sex; comorbid conditions at dialysis initiation: angina, previous coronary artery bypass graft, stroke, current smoker, diabetes, hypertension, lung disease, malignancy, myocardial infarction, peripheral vascular disease, pulmonary edema; distance from treatment facility; ethnicity), we observed no statistically significant change in the proportion of patients initiating dialysis early between the intervention and control arms (relative risk, 1.09; 95% CI, 0.91 to 1.30; P=0.37). With respect to patients’ acute dialysis initiations, we again saw no statistically significant change (relative risk, 1.00; 95% CI, 0.84 to 1.18; P=0.96).
In a post-hoc analysis among clinics that reported a high subjective ranking of the knowledge translation tools in a postintervention survey (12 of 27 clinics assigned to the intervention, 44%), we again did not find any statistically significant changes in the primary efficacy outcome (relative risk, 0.94; 95% CI, 0.66 to 1.34; P=0.73 for early initiation) or safety outcome (relative risk, 1.05; 95% CI, 0.92 to 1.19; P=0.46 for acute initiation) in the primary analysis, with similar findings in applying the difference-in-differences approach (Supplemental Material 3).
Discussion
In this national cluster randomized trial of 55 CKD clinics, our multifaceted knowledge translation intervention did not lead to an improvement in the proportion of early dialysis starts. There were no changes in the proportion of inpatient dialysis starts in the intervention clinics when compared with control clinics, and no differences were observed with respect to acute dialysis initiations as an inpatient. A secular trend toward deferring dialysis initiation after the publication of the IDEAL trial was observed in both treatment groups, and our intervention showed no effect in further lowering the rates of early initiation.19
The IDEAL trial, published in 2010, compared the efficacy and safety of an early versus late dialysis initiation strategy on patient outcomes and found no improvement associated with early initiation.4 Subsequently, clinical practice guidelines that incorporated findings from the IDEAL trial and multiple observational studies recommended an intent-to-defer approach to dialysis initiation, thereby advocating for a reduction in the proportion of early dialysis starts.5,20 However, substantial center-level variability in the timing of dialysis initiation remained post-IDEAL,21 and led us to design this multifaceted intervention to implement the evidence-based guideline.
To our knowledge, this is the first cluster-randomized trial to target the timing of dialysis initiation at a national scale and at the clinic level. Previous trials that have evaluated knowledge-translation interventions in nephrology and primary care CKD populations have achieved mixed results. In particular, studies targeting process outcomes such as measurement of albuminuria or prescription of renal angiotensin system inhibitors have been successful, whereas those targeting clinical outcomes such as CKD progression or hospitalizations have largely been negative.22–24 Similarly, interventions such as audit and feedback and academic detailing have been effective,14,25,26 whereas interventions that are primarily on the basis of decision support within the electronic health record have not.27,28 With these features in mind, we designed our multifaceted intervention to include audit and feedback, academic detailing, and provided several visual aids to enhance the uptake of the intent-to-defer strategy and decrease the proportion of patients who undergo early dialysis initiation.
There are several possible reasons why our intervention did not result in a detectable improvement in the proportion of early dialysis starts originating from the CKD clinics. In Canada, the publication of the IDEAL trial led to an immediate and sustained change in dialysis initiation practices without additional dissemination efforts.19 Early dialysis initiation peaked pre-IDEAL in 2009, and had a sustained meaningful decline from 2009 to 2015, which marked the start of our intervention. Our messages regarding the benefits from an intent-to-defer strategy may therefore have been perceived as confirmatory rather than practice changing, and most clinics continued to treat patients in a similar manner pre- or postintervention. In addition, ongoing scientific symposia at national meetings and provincial guidelines may have led to promotion of the intent-to-defer strategy in the control clinics, further limiting our ability to detect an improvement with our intervention. Further, it is possible that in some patients with sarcopenia, eGFR may be estimated as >10.5 ml/min per 1.73 m2 due to the association of creatinine with muscle mass and dietary protein intake, where the true eGFR may be <5 ml/min per 1.73 m2. This difference may become detectable as equations that use cystatin C become more common in clinical practice.29 Finally, it is also possible that our multifaceted intervention was ineffective for this treatment decision, and alternative intervention strategies such as changes in reimbursement may be needed to influence dialysis initiation.
Our study had some limitations. We intervened in structured multidisciplinary advanced CKD clinics only, and any patients without sustained nephrology follow-up or those followed in a nephrologist-only model of care would not have been affected with our intervention. Furthermore, the level of engagement with the intervention was inconsistent among clinics (with only 50% of clinics reporting a high level of engagement with the intervention), and we were unable to tailor the intervention to each clinic’s individual needs. In addition, our outcomes were collected from the CORR and were only on the basis of the proportion of patients who progressed to kidney failure and started dialysis. If the intervention led to a meaningful increase in the number of patients with an eGFR <10.5 ml/min per 1.73 m2 who stayed in the clinics without dialysis treatment, we would not have captured this outcome. We are working on obtaining and analyzing data from the CKD clinics, one province at a time. Dialysis data in Canada are available through a national registry, but clinic data require provincial approvals. We hope to report on these findings in 2021–22. Lastly, the cluster-level trial design of our study may have limited the power to detect small differences in the efficacy of the intervention.
Strengths of our study include its national scope, because we included all provinces except Quebec, and effectively randomized all Canadians with advanced CKD who were enrolled in nephrology-run interdisciplinary clinics. This trial therefore establishes a platform for future cluster randomized trials in Canada evaluating predialysis CKD practices. Furthermore, we conducted this trial using exclusively registry-based outcomes and were therefore able to reduce research-related costs substantially (<C$500,000 for the entire trial). As such, our trial again provides a template for investigators who wish to conduct large pragmatic randomized trials with a smaller budget.
In conclusion, a multifaceted knowledge translation intervention that included audit and feedback, academic detailing to improve provider knowledge, and patient-directed educational tools failed to reduce the proportion of early dialysis starts in patients with CKD followed in multidisciplinary clinics. Passive dissemination and rapid uptake of the IDEAL trial findings may have limited the ability of the intervention to have an additional effect on the timing of dialysis initiation.
Disclosures
A. Alam reports receiving honoraria from AstraZeneca, Janssen, Novo Nordisk, and Otsuka Pharmaceutical; reports having consultancy agreements with AstraZeneca, Bayer, Janssen Canada, Otsuka Canada; reports speakers bureau from Janssen Canada and Otsuka Canada; and other interests/relationships with Royal College International (Canada). A. Garg reports receiving partnership grant funding from Astellas Canada Inc. for a Canadian Institutes of Health Research funded grant in living kidney donation; reports being a scientific advisor to or member of the Editorial Board of Kidney International and American Journal of Kidney Diseases; and reports other interests/relationships as member of the Data Safety and Monitoring Board for an Investigator Initiated Trial Program Funded by GlaxoSmithKline, Medical Lead Role to Improve Access to Kidney Transplantation and Living Kidney Donation for the Ontario Renal Network (government funded agency located within Ontario Health). C. Rigatto reports consultancy agreements with AstraZeneca, Boehringer Ingelheim, Lilly, and Sanofi; reports receiving research funding from Sanofi; reports receiving honoraria from Boehringer Ingelheim, Lilly, and Sanofi; reports being a scientific advisor or member of the Canadian Journal of Kidney Health and Disease. G. Nesrallah reports receiving consulting fees from Amgen Inc. and Baxter Healthcare; reports consultancy agreements with AstraZeneca, Baxter Healthcare, and Otsuka; and reports receiving research funding from Baxter Healthcare. M. Sood has received educational funds from AstraZeneca and reports being a scientific advisor or member of the Editorial Board of CJASN, Editorial Board of American Journal of Kidney Disease, Editorial Board of Canadian Journal of Cardiology, Editor of Canadian Journal of Kidney Disease and Health, and Member of the ASN Highlights ESKD Team. N. Tangri reports receiving honoraria from AstraZeneca, Bayer, BI-Lilly, Boehringer Ingelheim Canada Ltd., Janssen, Otsuka Pharmaceutical, Pfizer, and Tricida Inc.; and has received research support from AstraZeneca Inc., Janssen, Otsuka and Tricida Inc.; reports consultancy agreements with Mesentech Inc., PulseData Inc., Renibus, and Tricida Inc.; ownership interest in Clinpredict Ltd., Mesentech Inc., PulseData Inc, Renibus, Tricida Inc.; and reports being a scientific advisor or member of Mesentech, Pulsedata Inc., Renibus, and Tricida Inc. P. Komenda reports receiving honoraria from AstraZeneca, Boehringer Ingelheim Canada, and Otsuka Pharmaceutical; reports being a member of the scientific advisory board for NxStage Medical Inc. and is the consultant Chief Medical Officer for Quanta Dialysis Technologies; reports consultancy agreements as Chief Medical Officer of Quanta Dialysis Technologies; reports an ownership interest in Quanta Dialysis Technologies; reports receiving research funding from AstraZeneca, Baxter, and NxStage; reports receiving honoraria from AstraZeneca, Boehringer Ingelheim, Janssen, Otsuka, and Quanta; and reports other interests/relationships with Chronic Disease Innovation Centre, Canadian Society of Nephrology Executive, and Seven Oaks Hospital Foundation Council. S. Kim has received support from Astellas Pharma Canada; and reports being a scientific advisor or member of Canadian Blood Services, Canadian Society of Transplantation, CORR, Health Canada, Scientific Registry of Transplant Recipients Technical Advisory Committee, and Organ Procurement and Transplantation Network/United Network for Organ Sharing Data Advisory Committee; and reports other interests/relationships through being Steering Committee member for the Canadian Transplant Forum (cosponsored by Astellas Pharma, Pfizer, and Alexion). S. Soroka reports receiving honoraria from AstraZeneca, Bayer, and Otsuka; and reports speakers bureau from AstraZeneca, Bayer, Jannsen, and Otsuka. T. Ferguson reports consultancy agreements with Clinpredict, Navdeep Tangri Medical Corporation, Quanta Dialysis Technologies, and Strategic Health Resources. All remaining authors have nothing to disclose.
Funding
This study was funded by a grant from Research Manitoba (Grant 1102).
Supplementary Material
Acknowledgments
The authors would like to thank Mr. Frank Ivis from the Canadian Institute for Health Information for help in data acquisition.
Footnotes
Published online ahead of print. Publication date available at www.jasn.org.
Supplemental Material
This article contains the following supplemental material online at http://jasn.asnjournals.org/lookup/suppl/doi:10.1681/ASN.2020091254/-/DCSupplemental.
Supplemental Material 1. Overview of randomization, intervention assignment, exclusion criteria, and follow-up for differences-in-differences analyses.
Supplemental Material 2. Baseline characteristic patients for differences-in-differences analyses.
Supplemental Material 3. Results of post-hoc analysis among clinics reporting a high subjective ranking of the provided knowledge translation tools (12 or 27 clinics assigned to the intervention) with GEE models in the 1 year after the KT intervention.
References
- 1.Clark WF, Na Y, Rosansky SJ, Sontrop JM, Macnab JJ, Glassock RJ, et al.: Association between estimated glomerular filtration rate at initiation of dialysis and mortality. CMAJ 183: 47–53, 2011 10.1503/cmaj.100349 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Harris A, Cooper BA, Li JJ, Bulfone L, Branley P, Collins JF, et al.: Cost-effectiveness of initiating dialysis early: A randomized controlled trial. Am J Kidney Dis 57: 707–715, 2011 10.1053/j.ajkd.2010.12.018 [DOI] [PubMed] [Google Scholar]
- 3.Rosansky SJ, Clark WF, Eggers P, Glassock RJ: Initiation of dialysis at higher GFRs: Is the apparent rising tide of early dialysis harmful or helpful? Kidney Int 76: 257–261, 2009 10.1038/ki.2009.161 [DOI] [PubMed] [Google Scholar]
- 4.Cooper BA, Branley P, Bulfone L, Collins JF, Craig JC, Fraenkel MB, et al.; IDEAL Study: A randomized, controlled trial of early versus late initiation of dialysis. N Engl J Med 363: 609–619, 2010 10.1056/NEJMoa1000552 [DOI] [PubMed] [Google Scholar]
- 5.Nesrallah GE, Mustafa RA, Clark WF, Bass A, Barnieh L, Hemmelgarn BR, et al.; Canadian Society of Nephrology: Canadian Society of Nephrology 2014 clinical practice guideline for timing the initiation of chronic dialysis. CMAJ 186: 112–117, 2014 10.1503/cmaj.130363 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Coronel F, Cigarrán S, Herrero JA: Early initiation of peritoneal dialysis in diabetic patients. Scand J Urol Nephrol 43: 148–153, 2009 10.1080/00365590802602903 [DOI] [PubMed] [Google Scholar]
- 7.Bao Y, 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 10.1001/archinternmed.2012.3020 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Grimshaw JM, Eccles MP, Lavis JN, Hill SJ, Squires JE: Knowledge translation of research findings. Implement Sci 7: 50, 2012 10.1186/1748-5908-7-50 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Chau EMT, Manns BJ, Garg AX, Sood MM, Kim SJ, Naimark D, et al.; Canadian Kidney Knowledge Translation and Generation Network (CANN-NET): Knowledge translation interventions to improve the timing of dialysis initiation: Protocol for a cluster randomized trial. Can J Kidney Health Dis 3: 2016, 10.1177/2054358116665257 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Tangri N, Stevens LA, Griffith J, Tighiouart H, Djurdjev O, Naimark D, et al.: A predictive model for progression of chronic kidney disease to kidney failure. JAMA 305: 1553–1559, 2011 10.1001/jama.2011.451 [DOI] [PubMed] [Google Scholar]
- 11.Moist LM, Richards HA, Miskulin D, Lok CE, Yeates K, Garg AX, et al.: A validation study of the Canadian organ replacement register. Clin J Am Soc Nephrol 6: 813–818, 2011 10.2215/CJN.06680810 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Juurlink D, Preyra C, Croxford R, Chong A, Austin P, Tu J, et al.: Canadian Institute for Health Information Discharge Abstract Database: A Validation Study, Ontario, Canada, Institute for Clinical Evaluative Sciences, 2006 [Google Scholar]
- 13.Sood MM, Manns B, Nesrallah G; Canadian Kidney Knowledge Translation, Generation Network: Using the knowledge-to-action framework to guide the timing of dialysis initiation. Curr Opin Nephrol Hypertens 23: 321–327, 2014 10.1097/01.mnh.0000445746.85502.42 [DOI] [PubMed] [Google Scholar]
- 14.Ivers N, Jamtvedt G, Flottorp S, Young JM, Odgaard-Jensen J, French SD, et al.: Audit and feedback: Effects on professional practice and healthcare outcomes. Cochrane Database Syst Rev 6[6]: CD000259, 2012 10.1002/14651858.CD000259.pub3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Jamtvedt G, Young JM, Kristoffersen DT, O’Brien MA, Oxman AD: Does telling people what they have been doing change what they do? A systematic review of the effects of audit and feedback. Qual Saf Health Care 15: 433–436, 2006, 10.1136/qshc.2006.018549 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D: A more accurate method to estimate glomerular filtration rate from serum creatinine: A new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med 130: 461–470, 1999 10.7326/0003-4819-130-6-199903160-00002 [DOI] [PubMed] [Google Scholar]
- 17.Spiegelman D, Hertzmark E: Easy SAS calculations for risk or prevalence ratios and differences. Am J Epidemiol 162: 199–200, 2005 10.1093/aje/kwi188 [DOI] [PubMed] [Google Scholar]
- 18.Dimick JB, Ryan AM: Methods for evaluating changes in health care policy: The difference-in-differences approach. JAMA 312: 2401–2402, 2014 10.1001/jama.2014.16153 [DOI] [PubMed] [Google Scholar]
- 19.Ferguson TW, Garg AX, Sood MM, Rigatto C, Chau E, Komenda P, et al.: Association between the publication of the initiating dialysis early and late trial and the timing of dialysis initiation in Canada. JAMA Intern Med 179: 934–941, 2019 10.1001/jamainternmed.2019.0489 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Chan CT, Blankestijn PJ, Dember LM, Gallieni M, Harris DCH, Lok CE, et al.; Conference Participants: Dialysis initiation, modality choice, access, and prescription: Conclusions from a Kidney Disease: Improving Global Outcomes (KDIGO) controversies conference. Kidney Int 96: 37–47, 2019 10.1016/j.kint.2019.01.017 [DOI] [PubMed] [Google Scholar]
- 21.Sood MM, Manns B, Dart A, Hiebert B, Kappel J, Komenda P, et al.; Canadian Kidney Knowledge Translation and Generation Network (CANN-NET): Variation in the level of eGFR at dialysis initiation across dialysis facilities and geographic regions. Clin J Am Soc Nephrol 9: 1747–1756, 2014 10.2215/CJN.12321213 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Peralta CA, Frigaard M, Rolon L, Seal K, Tuot D, Senyak J, et al.: Screening for CKD to improve processes of care among nondiabetic veterans with hypertension a pragmatic cluster-randomized trial. Clin J Am Soc Nephrol 15: 174–181, 2020 10.2215/CJN.05050419 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Tuot DS, McCulloch CE, Velasquez A, Schillinger D, Hsu CY, Handley M, et al.: Impact of a primary care CKD registry in a US public safety-net health care delivery system: A pragmatic randomized trial. Am J Kidney Dis 72: 168–177, 2018 10.1053/j.ajkd.2018.01.058 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Major RW, Brown C, Shepherd D, Rogers S, Pickering W, Warwick GL, et al.: The Primary-Secondary Care Partnership to Improve Outcomes in Chronic Kidney Disease (PSP-CKD) study: A cluster randomized trial in primary care. J Am Soc Nephrol 30: 1261–1270, 2019 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Chhina HK, Bhole VM, Goldsmith C, Hall W, Kaczorowski J, Lacaille D: Effectiveness of academic detailing to optimize medication prescribing behaviour of family physicians. J Pharm Pharm Sci 16: 511–529, 2013 10.18433/j3kk6c [DOI] [PubMed] [Google Scholar]
- 26.Fischer MA, Avorn J: Academic detailing can play a key role in assessing and implementing comparative effectiveness research findings. Health Aff (Millwood) 31: 2206–2212, 2012 10.1377/hlthaff.2012.0817 [DOI] [PubMed] [Google Scholar]
- 27.Manns B, Tonelli M, Culleton B, Faris P, McLaughlin K, Chin R, et al.; Alberta Kidney Disease Network: A cluster randomized trial of an enhanced eGFR prompt in chronic kidney disease. Clin J Am Soc Nephrol 7: 565–572, 2012 10.2215/CJN.12391211 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Abdel-Kader K, Fischer GS, Li J, Moore CG, Hess R, Unruh ML: Automated clinical reminders for primary care providers in the care of CKD: A small cluster-randomized controlled trial. Am J Kidney Dis 58: 894–902, 2011 10.1053/j.ajkd.2011.08.028 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Tangri N, Stevens LA, Schmid CH, Zhang YL, Beck GJ, Greene T, et al.: Changes in dietary protein intake has no effect on serum cystatin C levels independent of the glomerular filtration rate. Kidney Int 79: 471–477, 2011 10.1038/ki.2010.431 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.