Skip to main content

Foundational Trials

ATN

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Intensity of Renal Support in Critically Ill Patients with Acute Kidney Injury
  • Acronym: ATN (VA/NIH Acute Renal Failure Trial Network)
  • Year: 2008
  • Journal published in: The New England Journal of Medicine
  • Citation: Palevsky PM, Zhang JH, O’Connor TZ, et al; VA/NIH Acute Renal Failure Trial Network. Intensity of renal support in critically ill patients with acute kidney injury. N Engl J Med. 2008;359(1):7-20.

Context & Rationale

  • Background
    • Renal replacement therapy (RRT) for ICU acute kidney injury (AKI) had major practice variation in “dose” (intermittent frequency and small-solute targets; continuous effluent targets), despite high mortality and substantial resource use.
    • Single-centre trials had reported improved outcomes with higher-intensity strategies (e.g., higher-dose CVVH; daily intermittent dialysis), but these studies were small, heterogeneous, and potentially fragile to bias and local practice effects.12
    • Uncertainty persisted about whether increasing solute and fluid removal intensity modifies patient-centred outcomes (survival, kidney recovery) versus simply increases treatment-related complications.
  • Research Question/Hypothesis
    • In critically ill adults with AKI requiring RRT and at least one non-renal organ failure (or sepsis), does an intensive renal support strategy reduce 60-day all-cause mortality compared with a less-intensive strategy?
    • Secondary hypotheses addressed whether higher-intensity renal support improves kidney recovery, reduces RRT exposure, and mitigates non-renal organ failure.
  • Why This Matters
    • If “more dialysis” improved survival or recovery, it would justify additional circuit time, staffing, consumables, and risk.
    • If there were no benefit (or harm), the trial would support avoiding unnecessary intensity, improving safety, and re-focusing trials on timing, modality selection, and broader ICU-system care.

Design & Methods

  • Research Question: Whether an intensive, protocolised strategy of renal support reduces 60-day all-cause mortality versus a less-intensive strategy in ICU patients with AKI requiring RRT.
  • Study Type: Multicentre, parallel-group, randomised controlled trial (open-label), conducted in adult ICUs across VA and university-affiliated centres in the United States.
  • Population:
    • Setting: Critically ill adults in ICU with AKI requiring RRT and ≥1 non-renal organ failure (or sepsis).
    • Key inclusion criteria (protocol-defined): Adult ICU patient; AKI with need for RRT; at least one of: cardiovascular failure (SOFA CV 2–4), respiratory failure (mechanical ventilation), or sepsis; plus RRT indication(s) such as refractory fluid overload with pulmonary oedema, hyperkalaemia >6.5 mmol/L (or >6.0 with ECG changes), severe acidaemia (pH <7.2), and/or severe azotaemia (BUN >100 mg/dL).3
    • Key exclusion criteria (protocol-defined): Chronic kidney disease (baseline creatinine >2.0 mg/dL); kidney transplant; obstructive uropathy; rapidly progressive glomerulonephritis; advanced heart failure (NYHA IV); advanced liver disease; moribund/expected survival <28 days; DNR status; active participation in another intervention trial.3
  • Intervention:
    • Intensive strategy: Modality selected by haemodynamic status using cardiovascular SOFA score; intermittent therapies were delivered 6 days/week; continuous therapy targeted higher effluent.
    • Intermittent haemodialysis / SLED (when cardiovascular SOFA 0–2): 6 sessions/week; target urea clearance per session similar to control (Kt/V), but higher frequency and lower predialysis BUN achieved.
    • CVVHDF (when cardiovascular SOFA 3–4): prescribed effluent 35 ml/kg/hour (dialysate + replacement, excluding net ultrafiltration).
  • Comparison:
    • Less-intensive strategy: Same modality-selection rule (cardiovascular SOFA), but lower delivered frequency/effluent.
    • Intermittent haemodialysis / SLED: 3 sessions/week.
    • CVVHDF: prescribed effluent 20 ml/kg/hour (dialysate + replacement, excluding net ultrafiltration).
  • Blinding: Unblinded (pragmatic delivery of dialysis strategies); primary outcome (mortality) was objective, but open-label delivery could influence co-interventions and withdrawal decisions.
  • Statistics: Planned sample size 1164 to detect a 10% absolute reduction in 60-day mortality (from 55% to 45%) with 90% power at α=0.05; target increased for anticipated 10% loss; primary analysis intention-to-treat using conditional logistic regression stratified by site and cardiovascular SOFA group; 6 interim analyses planned.
  • Follow-Up Period: Mortality to day 60; kidney recovery assessed to day 28; organ failure-free days assessed to day 14; RRT exposure through day 28; discharge status to day 60.

Key Results

This trial was not stopped early. Enrolment was 1124 participants (vs 1164 planned), with prespecified interim monitoring; no efficacy or harm boundary triggered cessation.

Outcome Intensive strategy Less-intensive strategy Effect p value / 95% CI Notes
Death from any cause by day 60 (primary) 302/563 (53.6%) 289/561 (51.5%) OR 1.09 95% CI 0.86 to 1.40; P=0.47 No mortality benefit with higher intensity.
In-hospital death 277/563 (49.2%) 265/561 (47.2%) OR 1.08 95% CI 0.84 to 1.38; P=0.60 Consistent with primary outcome.
Complete recovery of kidney function by day 28 85/553 (15.4%) 102/555 (18.4%) OR 0.81 95% CI 0.59 to 1.13; P=0.24 Recovery definitions incorporated baseline creatinine; missing data excluded from denominator.
RRT-free days through day 28 (mean ± SE) 6.0 ± 0.4 7.0 ± 0.4 Mean difference −0.9 days 95% CI −1.9 to 0.1; P=0.07 Days alive and free of RRT; deaths counted as 0 days.
Hospital-free days through day 60 (mean ± SE) 11.0 ± 0.7 13.0 ± 0.7 Mean difference −1.9 days 95% CI −3.9 to 0.0; P=0.053 Borderline difference, not statistically significant at α=0.05.
ICU-free days through day 60 (mean ± SE) 18.7 ± 0.9 20.1 ± 0.9 Mean difference −1.5 days 95% CI −4.0 to 1.0; P=0.25 No meaningful separation in ICU-free days.
Discharged to home and off dialysis by day 60 35/563 (6.2%) 46/561 (8.2%) OR 0.74 95% CI 0.46 to 1.17; P=0.19 Low absolute rates in this very sick cohort.
Hypophosphataemia 99/563 (17.6%) 61/561 (10.9%) OR 1.75 95% CI 1.23 to 2.49; P=0.002 More electrolyte depletion with intensive dosing.
Hypokalaemia 136/563 (24.1%) 100/561 (17.8%) OR 1.46 95% CI 1.08 to 1.98; P=0.01 Clinically relevant safety signal.
Hypotension requiring intervention during intermittent dialysis 178/472 (37.7%) 140/467 (30.0%) OR 1.41 95% CI 1.08 to 1.82; P=0.01 Assessed among those receiving intermittent modalities.
  • Clear separation in delivered intensity was achieved, yet without improvement in survival or renal recovery.
  • Secondary outcomes (RRT-free, ICU-free, hospital-free days) showed no statistically significant benefit of higher intensity, with some point estimates numerically favouring less-intensive therapy.
  • Harms increased with intensive strategies, particularly electrolyte depletion (hypophosphataemia, hypokalaemia) and intradialytic hypotension requiring intervention.

Internal Validity

  • Randomisation and Allocation:
    • Central interactive randomisation; stratified by site and cardiovascular SOFA category (0–2 vs 3–4), and by oliguria within intermittent-modality strata.
    • Allocation concealment was plausibly robust until randomisation; selection bias mitigated by centralised assignment.
  • Drop out / exclusions (post-randomisation):
    • Study therapy withdrawn: 18/563 (3.2%) intensive vs 11/561 (2.0%) less-intensive (reasons included discharge, withdrawal of consent, clinician decision).
    • Vital status not ascertained for 5 participants (2 intensive; 3 less-intensive); analysed as alive by protocol.
    • Kidney recovery analyses had missing baseline creatinine for a minority (denominators 553 and 555 for recovery categories).
  • Performance/Detection Bias:
    • Unblinded delivery could influence co-interventions (fluid management, nutrition, drug dosing) and decisions about continuing RRT.
    • Primary endpoint was objective (all-cause mortality), reducing detection bias risk for the main outcome.
  • Protocol Adherence:
    • Intermittent therapies: missed treatments 1.9% intensive vs 1.1% less-intensive; extra treatments 0.5% vs 1.5%.
    • Delivered dose: intermittent Kt/V per session similar (≈1.2–1.3), with separation primarily via frequency; continuous effluent delivery reached ~89% (intensive) vs ~95% (less-intensive) of prescribed.
  • Baseline Characteristics:
    • Groups were well balanced; cohort was severely ill (mean APACHE II 26.4 ± 7.3; mean SOFA 14.5 ± 3.7; 80.6% mechanically ventilated; 63.0% sepsis; 78.0% oliguria).
    • 64.5% received up to 24 hours of RRT prior to randomisation, indicating the comparison tested “intensity strategy after initiation” rather than timing of initiation.
  • Heterogeneity:
    • Prespecified subgroup analyses showed no convincing effect modification (interaction thresholds prespecified; no subgroup clearly favoured intensive therapy).
    • Examples: cardiovascular SOFA 0–2 OR 1.33 (95% CI 0.93 to 1.91) vs SOFA 3–4 OR 0.93 (0.66 to 1.29); interaction P=0.15; oliguria yes OR 1.04 (0.79 to 1.37) vs no OR 1.31 (0.77 to 2.21); interaction P=0.45.
  • Timing:
    • Randomisation occurred at (or shortly after) meeting RRT criteria; however, prior receipt of RRT in ~two-thirds could attenuate a true “early high-intensity” effect.
  • Dose:
    • The intensive strategy materially increased delivered clearance exposure (see separation below), but the less-intensive strategy still represented a relatively “adequate” modern dose (e.g., CVVHDF 20 ml/kg/hour), potentially limiting marginal gains.
  • Separation of the Variable of Interest:
    • Intermittent arm intensity separation: 5.4 ± 1.4 vs 3.0 ± 0.7 treatments/week; interval 1.1 ± 0.4 vs 2.0 ± 0.4 days; predialysis BUN 45 ± 25 vs 70 ± 33 mg/dL.
    • Continuous arm intensity separation: delivered effluent 35.8 ± 6.4 vs 22.0 ± 6.1 ml/kg/hour; achieved BUN 33 ± 20 vs 47 ± 25 mg/dL.
  • Key Delivery Aspects:
    • Modality was algorithm-driven by haemodynamic status (cardiovascular SOFA), aligning with pragmatic ICU practice and limiting clinician “cherry-picking” of modality by arm.
    • Discontinuation guidance used timed urine collections and urea/creatinine clearance thresholds (stop RRT if clearance >20 ml/min; restart if <12 ml/min), supporting standardised stopping rules.
  • Outcome Assessment:
    • Mortality objective; kidney recovery definitions operationalised using serum creatinine relative to baseline (complete: ≤0.5 mg/dL above baseline; partial: >0.5 mg/dL above baseline; no recovery: dialysis-dependent or missing).
  • Statistical Rigor:
    • Intention-to-treat analysis with stratified modelling; prespecified interim monitoring.
    • Enrolment slightly below target (1124 vs 1164 planned), reducing power for small absolute mortality differences.

Conclusion on Internal Validity: Overall, internal validity appears strong for the primary endpoint, supported by central randomisation, objective mortality assessment, high protocol adherence, and clear dose separation; the key threats are open-label delivery and prior receipt of RRT in many participants, which could dilute an effect of earlier or very-high-intensity initiation.

External Validity

  • Population Representativeness:
    • Represents a very high-acuity ICU AKI population (high rates of mechanical ventilation, sepsis, and oliguria) in predominantly VA/university centres.
    • Exclusion of advanced chronic kidney disease (baseline creatinine >2.0 mg/dL), transplant recipients, and select primary renal diagnoses limits applicability to these groups.
  • Applicability:
    • Modality-switching guided by haemodynamics is broadly aligned with real-world ICU decision-making.
    • The specific intermittent schedule (6 vs 3 sessions/week) and CVVHDF dosing targets translate best to settings with reliable delivery capacity; generalisability may be reduced where downtime, staffing constraints, or consumable limitations materially reduce delivered effluent/dose.

Conclusion on External Validity: Findings are highly applicable to severely ill ICU patients with established indications for RRT in well-resourced systems, but are less generalisable to earlier-stage AKI, patients with advanced pre-existing kidney disease, and resource-limited environments where delivered dose may differ substantially from protocol targets.

Strengths & Limitations

  • Strengths:
    • Large, multicentre randomised trial in a population at high risk of death and dialysis dependence.
    • Protocolised, haemodynamic rule-based modality allocation with the same algorithm in both arms.
    • Clear achieved separation in biochemical and delivered-dose metrics (BUN and effluent/frequency).
    • Objective primary endpoint; systematic capture of clinically relevant complications.
  • Limitations:
    • Open-label design; potential for co-intervention and withdrawal biases, particularly for non-mortality outcomes.
    • Many patients received RRT before randomisation, so the trial does not directly address “early high-intensity” initiation.
    • Enrolment below planned sample size reduces power for small treatment effects.
    • Integrated strategy across modalities answers a pragmatic clinical question, but complicates mechanistic inference about modality-specific dosing effects.

Interpretation & Why It Matters

  • Clinical practice
    • Escalating RRT intensity beyond a “conventional/adequate” dose (CVVHDF 20 ml/kg/hour; intermittent 3 sessions/week) did not improve survival or renal recovery, supporting avoidance of routine “more is better” dialysis in ICU AKI.
    • Higher intensity increased clinically important complications (electrolyte depletion; intradialytic hypotension), reinforcing a safety–resource trade-off.
  • Mechanistic inference
    • In severe ICU AKI, death appears dominated by underlying multi-organ failure biology and treatment limitations rather than by incremental increases in small-solute clearance once an adequate dose threshold is met.
  • Trial design legacy
    • ATN helped establish contemporary dosing “guardrails” and a standardised approach to delivered dose accounting and safety monitoring, which informed subsequent trials and guideline thresholds.

Controversies & Subsequent Evidence

  • Interpretation of “negative” dosing trials:
    • Editorial commentary emphasised that once an adequate dialysis dose is delivered, escalating intensity may not overcome the mortality risk imposed by multi-organ failure—and may increase complications.4
    • Critical appraisal argued against “therapeutic nihilism”, highlighting the importance of ensuring adequate delivered dose, accounting for downtime and dilutional effects, and separating dose questions from timing and patient selection questions.5
  • Was the control arm already “high”?
    • The less-intensive strategy used CVVHDF 20 ml/kg/hour and intermittent Kt/V targets consistent with modern “adequate” dosing; if benefit exists only at low delivered doses, ATN would not detect it (a floor effect).
  • Modality-mixing and pragmatic switching:
    • ATN’s haemodynamic algorithm made the study clinically pragmatic, but it means the causal contrast is a “strategy” rather than a pure modality- or solute-clearance experiment.
  • Subsequent RCTs:
    • Large multicentre evidence in continuous RRT similarly found no mortality benefit from higher-intensity CRRT dosing, aligning with ATN’s conclusion that routine dose escalation does not improve patient-centred outcomes once adequate dosing is achieved.

Summary

  • In severe ICU AKI requiring RRT, an intensive renal support strategy did not reduce 60-day all-cause mortality versus a less-intensive strategy.
  • Kidney recovery and key time-to-event surrogates (RRT-free, ICU-free, hospital-free days) were not improved with higher intensity.
  • The trial achieved clear separation in delivered dose (IHD frequency and CVVHDF effluent rate), strengthening confidence that a true “dose intensification” contrast was tested.
  • Higher intensity increased clinically important harms, particularly electrolyte depletion (hypophosphataemia, hypokalaemia) and intradialytic hypotension requiring intervention.
  • ATN helped define contemporary “adequate dose” practice and shifted attention towards timing, patient selection, and broader ICU multi-organ failure biology rather than routine dose escalation.

Further Reading

Other Trials

Systematic Review & Meta Analysis

Observational Studies

Guidelines

Notes

  • Further Reading is limited here to sources with DOI-confirmable linking available in the source set; additional recent observational cohorts and guideline documents exist but are not listed to avoid unverified bibliographic details.

Overall Takeaway

ATN is a landmark trial because it convincingly tested an ICU-relevant, strategy-level question—whether increasing RRT intensity improves outcomes—while achieving clear delivered-dose separation in a severely ill cohort. Its neutral efficacy results, coupled with increased treatment-related complications, shifted practice away from routine “dose escalation” and toward delivering an adequate, safe dose while prioritising patient selection, timing questions, and multi-organ failure management.

Overall Summary

  • In ICU AKI requiring RRT, “more intensive” renal support did not reduce 60-day mortality and increased adverse metabolic and haemodynamic events.

Bibliography