#NephJC Chat
Tuesday, June 24th, 2025, 9 pm Eastern on Bluesky
Circulation. 2025 Jun 7. doi: 10.1161/CIRCULATIONAHA.125.075403. Online ahead of print.
A Nationwide Factorial Randomized Trial of Electronic Nudges to Patients with Chronic Kidney Disease and Their General Practices for Increasing Guideline-Directed Medical Therapy: The NUDGE-CKD Trial
Kristoffer Grundtvig Skaarup, Niklas Dyrby Johansen, Lisbet Brandi, Morten Kofod Lindhardt, Jesper N Bech, My Svensson, Tilde Kristensen, Anne Daugaard Thuesen, Majbritt Grønborg Knudsen, Jan Dominik Kampman, Mads Hornnum, Birgitte Ørts, Daniel Modin, Mats C H Lassen, Kira Hyldekær Janstrup, Brian L Claggett, Muthiah Vaduganathan, Ankeet S Bhatt, Harriette G C Van Spall, Jens Ulrik Stæhr Jensen, Faiez Zannad, Scott D Solomon, Anne Møller, Rikke Borg, Henrik Birn, Ditte Hansen, Tor Biering-Sørensen
PMID: 40481660
Introduction
Chronic kidney disease (CKD) remains underdiagnosed and undertreated despite its rising global burden. While low awareness is often blamed for delayed detection and progression, improving outcomes requires more than education and demands sustainable, scalable solutions that drive action (Francis et al, Nat Rev Nephrol 2020). Across healthcare systems CKD care is overwhelmingly delivered by primary care: in Denmark only 3% of patients with eGFR <60ml/min, and just 16% with eGFR <30 are referred to a nephrologist within a year (Kampmann JD, et al, Clin Kidney J 2022). Similar gaps are seen internationally: over half of US patients with advanced CKD remain under exclusive primary care management (Dharod A, et al, Am J Mang Care 2020), and in Japan fewer than 2% of screen-detected CKD patients initiate follow-up with a physician (Yamada Y, et al, J Epidemiol Community Health 2019). Referral is typically reserved for younger, sicker, and more complex patients, leaving the majority of CKD care, including the initiation of disease-modifying therapies, to general practitioners.
Yet the story for CKD is changing. Over the past few years, an unprecedented wave of landmark trials have reshaped the therapeutic landscape, from RASi, flozins (SGLT2i renal outcomes summary), to mineralocorticoid receptor antagonists (MRAs) and GLP-1 receptor agonists, all of which have demonstrated improved renal and cardiovascular outcomes, fundamentally transforming our approach to CKD care (KDIGO 2024, summary).
Despite advances in clinical guidelines, integrating these therapies into routine practice remains a major challenge, particularly in primary care, where the majority of CKD patients are managed. The time of treating CKD with a single therapy is fading. CKD is a heterogeneous condition, shaped by diverse pathophysiological mechanisms, fluctuating risks, and evolving treatment priorities. As the number of effective interventions expands, so does the complexity of clinical decision-making. More evidence does not automatically translate into better care. However, prescription rates of guideline-recommended therapies for CKD (like RASi and SGLT2i) remain low across all settings. In the US, only 17.8% of patients with CKD and diabetes initiated ACEi/ARBs, and just 2.5% received SGLT2i, with high discontinuation rates. (Fried L, et al, Nephrol Dial Transplant, 2023) In the UK, only 17% of patients with a clear indication for SGLT2i were treated- use was especially low in those without diabetes or with more advanced CKD (Forbes AK et al, EClinicalMedicine 2024). In Denmark, just 14% of patients with CKD managed by GP were prescribed SGLT2i, despite high comorbidity (Lindhardt M, et al, BMC Prim Care 2025).
In this sense, nephrology finds itself in a therapeutic adolescence: armed with powerful tools, but still searching for coherence, integration, and effective implementation. This persistent gap in implementation has driven interest in behavioral science-based approaches such as “nudging” to improve clinical decision-making (Last et al, BMJ 2021). Here is how Briana Last described “Nudges”
“
Research from economics, cognitive science and social psychology have converged on the finding that human rationality is ‘bounded’. The intractability of certain decision problems, constraints on human cognition and scarcity of time and resources lead individuals to employ mental shortcuts to make decisions. These mental shortcuts, often called heuristics, are strategies that overlook certain information in a problem with the goal of making decisions more quickly than more deliberative methods. While heuristics can often be more accurate than more complex mental strategies, they can also lead to errors and suboptimal decisions. Researchers have discovered interventions to harness the predictable ways in which human judgement is biased to improve decisions. These interventions, known as ‘nudges,’ reshape the ‘choice architecture,’ or the way options are presented to decision-makers, to optimise choices. Nudges have been applied to retirement savings, organ donation, consumer health and wellness, and climate catastrophe mitigation demonstrating robust effects.
.”
Nudges have been successfully employed across various medical domains, particularly in cardiovascular care, antibiotic stewardship, and preventive services. However, their effectiveness depends on timing, personalization, and integration into clinical workflows (Yoong et al, BMJ 2021).
In CKD nudging utility is uncertain. In a Japanese randomized trial, including over 4,000 individuals flagged by a national screening program, a behaviorally informed letter modestly increased nephrology follow-up (+3.9%, p=0.02), but had no effect on downstream clinical outcomes, including eGFR, proteinuria, or blood pressure at one year (Fukuma et al, JASN 2022).
The Study
To explore whether behavioral strategies can help close this implementation gap in CKD care, the NUDGE trial was launched. This study evaluated whether electronic nudges could increase the uptake of guideline therapies, specifically RASi and SGLT2i among patients with CKD.
Methods
NUDGE-CKD was a pragmatic, registry-based, 2x2 factorial randomized implementation trial embedded in the Danish healthcare system. Patients and general practices were independently randomized to receive either a nudge letter, or usual care. All analyses were conducted at the individual patient level.
The trial tested two distinct nudges- one directed at patients with CKD and the other at their GPs- to increase uptake of guideline-directed medical therapy (GDMT), specifically RASi or SGLT2i. Importantly, randomizations were independent: each patient and each practice could be in any combination of intervention and control arms.
The intervention was rolled out nationally on August 19, 2024, coinciding with the release of new national CKD guidelines. Follow-up ended on February 19th, 2025.
Population
The trial included:
All Danish citizens with CKD, defined as a hospital-coded CKD diagnosis, both outpatient and inpatient, within the past 5 years
Patients had to be enrolled in the digital post system to receive the patient-facing intervention (approx. 95% of the population)
For the provider-facing intervention, patients had to be listed with a regular GP
To assed broader practice-level effects, a secondary population comprising all adult patients registered with the randomized general practices was included. This cohort- 4.7 million individuals, covering approximately 80% of Denmark's population- served as a key comparator for exploratory analyses of GP-level prescribing behavior.
Figure 2. Overview of secondary population, from Skaarup KG, et al, Circulation 2025 (supplement)
Interventions
Patient letter: a single, standardized electronic letter emphasizing GDMT benefits, the updated Danish Society of Nephrology guidelines, and encouraging contact with their GP.
GP letter: A brief guideline-based summary with CKD definitions per KDIGO 2024, and hyperlink with full recommendations.
Letters were crafted by a multidisciplinary steering group and refined through input from patients and GPs. Initially, a second patient letter had been planned, but it was dropped due to concerns about overburdening practices.
Delivery mechanism
All letters were sent via digital post, Denmark’s secure national e-communication platform for official correspondence. Letters to GPs were sent to their practice’s official digital inbox.
Data sources
Outcomes were drawn entirely from Danish national administrative registries. These included:
Diagnoses: National Patient Registry (ICD-10 codes)
Prescriptions: National Prescription Registry
Procedures/Labs: Other national registries
Endpoints
The primary endpoint was the filled prescription of RASi or SGLT2i within 6 months post-intervention.
Secondary endpoints included individual drug classes, new users only, and time-to-prescription outcomes.
Sample size and power
As this was a nationwide registry-based trial including all eligible individuals, the sample size was fixed by design. Rather than using power calculation to determine sample size, the author estimated that the available cohort provided 80% power to detect a 1.8 percentage point absolute difference, assuming 65% baseline uptake for the patients level intervention (N= 22,617).
For the provider-level intervention (28,069 patients across 1,540 practices), the same assumptions were used, incorporating an intra-cluster correlation coefficient (ICC) of 0.0059 to account for within-practice similarity in prescribing behavior. This low ICC was later confirmed using a mixed-effects logistic regression model. This approach is standard in population-wide trials with fixed enrollment.
Because patient- and provider-level randomizations were conducted independently, clustering was accounted for only in analyses related to the provider-level intervention.
Ethics and oversight
The trial was approved by relevant Danish authorities, with a waiver of informed consent (communication-only intervention, no infringement on autonomy). While formal blinding was not feasible, patients and practices in the control arms were not informed of their allocation. Letters were co-signed by nephrologists from all five Danish regions.
Funding
The trial received no external funding. The lead investigator was supported by a grant from the Danish Cardiovascular Academy, funded by the Novo Nordisk Foundation and the Danish Heart Foundation. The funding didn’t influence the trial’s design, conduct, analysis, or reporting.
Results
Patients and General Practices
A total of 28,388 adults met the CKD-specific inclusion criteria, with 22,617 (79.7%) not exempt from the digital letter system. They were randomized to receive either an electronic nudge letter (n=11,223) or usual care (n=11,394) in the patient-level intervention. For the provider-level intervention, 28,069 (98.9%) adults were linked to a general practice eligible for randomization, spread across 1,540 general practices (with a median of 14 participants per practice). These practices were randomized to receive either the provider-facing intervention letter (n=774 practices, 13,959 patients) or usual care (n=766 practices, 14,110 patients).
Figure 2. Study flowchart for each intervention, from Skaarup KG, et al, Am J Heart 2025
Baseline characteristics were balanced across the randomization groups for each intervention, comprising a predominantly elderly population with a male predominance of approximately 60%. Nearly 40% had an eGFR <30, and 60% had elevated albuminuria (A2). The most common comorbidity was hypertension, followed by diabetes and chronic cardiovascular disease. Approximately 60% were already receiving RASi, and a quarter were on an SGLT2i (Table 1).
Comparisons between included and excluded individuals for each intervention are provided in the supplement. Notably, those who were excluded were older (median, 80 vs 71) and had higher rates of chronic cardiovascular disease (60% vs 48%), heart failure (18% vs 13%), and ischemic heart disease (22% vs 17%). (Table S5).
Primary and Secondary Outcomes
Primary and Secondary Endpoints for the Patient-Level Intervention
During the 6-month follow-up period, 65.1% of patients randomized to receive the patient-facing letter met the primary composite endpoint of filled prescription for RASi or SGLT2i, compared with 65.9% in the usual care group. This corresponds to a non-significant difference of -0.79 percentage points (95% CI, -2.03 to 0.45; P=0.21) (Figure 2). No interaction was observed with the provider-level intervention (p for interaction = 0.85).
Figure 2. Results for Primary and Secondary Endpoints for the Patient-Level Intervention, from Skaarup KG, et al, Circulation 2025
No significant between-group differences were observed for the individual components of the primary composite endpoint or among new users. Proportions with a filled prescription for RASi and SGLT2i showed opposite numerical directions of effect; however, no significant differences were observed between groups for the individual components of the primary composite endpoint or among new users only. Similar effect estimates were noted in time-to-event analyses of new prescriptions. No evidence of treatment heterogeneity was found across pre-specified subgroups (Supplement, Figure S3).
Primary and Secondary Endpoints for the Provider-Level Intervention
The provider-level intervention was also not significant, with 63.9% of individuals listed with general practices assigned to the provider-facing letter meeting the primary composite endpoint, compared to 64.4% in the usual care group, resulting in a difference of -0.49 percentage points (95% CI, -1.64 to 0.66; P=0.41) (Figure 4).
Figure 4. Results for Primary and Secondary Endpoints for the Provider-Level Intervention, from Skaarup KG, et al, Circulation 2025
Results were similar for the secondary endpoints, and comparable effect estimates were observed in time-to-event analyses of new prescriptions as well. Subgroup analyses did not indicate treatment effect heterogeneity for the primary endpoint (Supplement, Figure S4).
Participant Perception
A total of 4,646 respondents (41.3%) completed the self-administered electronic acceptability questionnaire in the patient-level intervention group (Supplement, Appendix III: Participant acceptability questionnaire). The results indicated that 80.4% of respondents read the intervention letter, but even fewer (23.5%) reported taking action as a result. Most of these individuals (56.6%) contacted their nephrology hospital department, while a slightly smaller number (42.5%) consulted their general practitioner. Among those who took action, 12.8% reported changes in their treatment. Overall, 81.5% of respondents expressed a positive attitude towards receiving similar updates about CKD treatment in the future.
Assessment in the Entire Patient Population of Randomized General Practices
The secondary population, comprising all adult citizens on the patient lists of randomized practices, included 4,772,890 patients (Supplement, Table S6). The median number of adult patients on patient lists of randomized practices was 2,654 (IQR, 1,526.5-3,944). Data on filled prescriptions of GDMT for CKD across randomization groups in this population during follow-up are presented in Table 3. Proportion of prescriptions of GDMT did not differ between patients with a general practice randomized to receive the intervention letter and those with a practice that did not.
Table 3. GDMT prescription rates in the entire patient panel of randomized general practices during follow-up, from Skaarup KG, et al, Circulation 2025
Discussion
The NUDGE-CKD trial was a nationwide, 2x2 factorial, pragmatic, randomized implementation trial that may have relied too heavily on a "pragmatic trial" approach, expecting too much from its participants through a single electronic “nudge” intervention. It implemented an electronic nudge letter intervention on GDMT in CKD for individuals with CKD and their general practices to increase GDMT uptake over 6 months compared to usual care (i.e., no electronic “nudge” letter). Neither group demonstrated a meaningful effect, nor were there signs of changes in CKD-related clinical practice among the randomized general practices. The novelty of this trial lies in its use of a digitally delivered, low-cost strategy on a national scale for a common and significant disease. Although it was a promising concept that required minimal effort, it ultimately failed to make any difference.
The electronic nudge letter sent to participants included information stating that treatment guidelines for CKD had been updated, emphasized the potential benefits of new therapies, and encouraged individuals to contact their healthcare provider at their convenience. However, the letter was neither personalized nor tailored to include details about the patients' most recent eGFR, UACR, diabetes status, and prescription history. It could have explicitly outlined which treatments the recipient might be eligible for based on established guideline criteria. This lack of actionable and specific recommendations might have contributed to clinical inertia, and did not convey the urgency of being prescribed GDMT. Additionally, general practitioners may lack the necessary knowledge or expertise to apply GDMT, hindering their ability to confidently discuss these changes with patients even when patients initiate the discussion. However, the authors argued that offering a more detailed or elaborate letter might be perceived as overly intrusive or burdensome to patients or providers, giving the impression that the care delivered may be lacking. This strategy would also require much greater manpower to review records and tailor individual letters.
The “nudge” strategy has proven effective in another study, the NOTIFY-1 project (Incidental Coronary Calcification Quality Improvement Project) (Sandhu et al, Circulation 2023), where patients and primary care providers were alerted to incidental coronary artery calcification found on previous chest CT scans. The notification included a patient-specific image of CAC and guideline recommendations for the use of statins. However, the NOTIFY-1 used a validated deep learning algorithm with radiologist confirmation, which would be difficult to scale or offer at a low cost.
A “nudge” strategy can be more effective when it is personalized and empowers the recipient to take actionable steps; however, this can gradually shift from being a subtle, small intervention to a more “change-oriented approach.” There are multiple models describing change theory and management that can be adapted for use in healthcare, such as Kotter’s 8-Step Change Model and Lewin’s Unfreezing/Moving/Refreezing Model (Harrison et al, J Healthc Leadersh, 2021). Both models generate a sense of urgency for why change is necessary, communicate this need, implement an intervention that produces the desired change, and establish the change as the new normal. Failure to implement change in healthcare can be attributed to poor planning, unmotivated staff, ineffective communication, and the complexity of widespread changes (Nelson-Brantley and Ford, J Adv Nurs 2017). Thus, increasing GDMT uptake requires not only behavioral nudges but also a deeper understanding of the multifaceted forces that promote or hinder change.
Multi-component strategies and resource-intensive approaches may enhance the adoption of GDMT with greater confidence. An example of this is a recent cluster-randomized trial conducted in the US, involving 141 primary care clinics managing 11,182 patients with the kidney-dysfunction triad (CKD, type 2 diabetes, and hypertension)(Vazquez et al, NEJM 2024). These patients received an intervention that utilized a personalized algorithm to identify them, along with practice facilitators to assist providers in delivering guideline-based interventions or continuing with usual care. Although the primary goal of reducing hospitalizations at one year was not achieved, a random review of a 10% subsample revealed higher rates of RASi prescriptions (11% vs. 6%). Additionally, another cluster-randomized clinical trial involving 43 US cardiology clinics aimed to increase the prescription of three evidence-based therapies (high-intensity statins, ACEi or ARB, and SGLT2i and/or GLP1-agonist) among adults with type 2 diabetes and atherosclerotic cardiovascular disease (Pagidipati et al., JAMA 2023). The intervention was delivered by assessing local barriers, developing care pathways, coordinating care, educating clinicians, reporting data back to the clinics, and providing participants with tools. This coordinated, multifaceted intervention successfully achieved its primary goal of increasing the prescription of three groups of evidence-based therapies (37.9% vs 14.5%, 23.4% difference).
Nudging should be viewed as a valuable co-factor—similar to a catalyst in a chemical reaction—that promotes change when other conditions are favorable. The success of nudging relies on its timing, integration, and contextual relevance. A nudge embedded in a clinical encounter—such as an EHR alert during a patient visit—may prompt action more effectively than a generic message sent without context. Nudges also work best when aligned with clinicians’ workflows and designed to reduce alert fatigue. Customizing nudges for the clinical setting, iterating based on user feedback, and integrating them into a broader support system are all key to maximizing their effectiveness. Ultimately, nudges lower the barriers to initiating change, but real and timely transformation occurs when they are combined with comprehensive, system-wide strategies and an organizational commitment to quality improvement.
Limitations:
The NUDGE-CKD trial, while innovative in scope and design, had several limitations that may explain its neutral findings. Conducted in Denmark, a country with universal healthcare and few access barriers, the baseline system might already support high levels of care engagement, which could limit the additional impact of the intervention. Furthermore, although over 95% of the population is reached through the national digital mail system, it remains unclear how many participants actually read or absorbed the nudge messages. This lack of insight into user engagement creates a significant blind spot. Additionally, the trial did not gather qualitative data to understand why patients or providers did not respond to the nudges, leaving unanswered questions about behavioral resistance, clinical inertia, or environmental barriers. Future research should use mixed-methods approaches to explore these factors and determine how nudges can be improved and tailored to different healthcare systems, especially those with weaker infrastructure or more significant disparities in care.
Conclusion
The NUDGE-CKD trial illustrates a core challenge in implementation: a cue without clarity or consequence rarely drives action. Like Pavlov’s dogs hearing a bell with no food, both patients and providers received a signal, but one not tied to an immediate, tangible response.
In the context of CKD, care is fragmented, decisions are complex, implementation demands more than information - it requires timing, personalization, and integration. Nudges may help, but without context and specificity, they risk becoming background noise in a system already tuned to inertia.