• 1.

    Fu EL, et al. Timing of dialysis initiation to reduce mortality and cardiovascular events in advanced chronic kidney disease: Nationwide cohort study. BMJ 2021; 375:e066306. doi: 10.1136/bmj-2021-066306

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 2.

    Cooper BA, et al. A randomized, controlled trial of early versus late initiation of dialysis. N Engl J Med 2010; 363:609619. doi: 10.1056/NEJMoa1000552

  • 3.

    Fu EL. @FuEdouard: “Why did observational studies find increased survival for late dialysis initiation, whereas the IDEAL RCT found no difference?Twitter, 6:30 p.m., Nov. 30, 2021. https://twitter.com/FuEdouard/status/1465840754820059146

    • PubMed
    • Search Google Scholar
    • Export Citation

Early Dialysis Improves Survival—but Is the Tradeoff Worth It?

Timothy O’Brien
Search for other papers by Timothy O’Brien in
Current site
Google Scholar
PubMed
Close
Full access

For patients with advanced chronic kidney disease (CKD), early dialysis initiation—at an estimated glomerular filtration rate (eGFR) of 15–16 mL/min/1.73 m2—leads to modest reductions in mortality and cardiovascular events, reports a study in The BMJ (1).

“However, to reach the maximum survival benefit, patients would need to start dialysis up to 4 years earlier,” comments lead author Edouard Fu, PhD, a research fellow at the Division of Pharmacoepidemiology and Pharmacoeconomics at Brigham and Women's Hospital and Harvard Medical School, Boston, Massachusetts.

The conclusions are consistent with the sole previous randomized trial of dialysis initiation times—and support current guideline recommendations on dialysis initiation. Fu and colleagues write: “Our findings provide novel evidence on the optimal timing of dialysis initiation and show that even with maximum eGFR separations, the range of plausible effects is likely to be small.”

Analysis “explicitly mimics” clinical trial of dialysis initiation times

The sole randomized trial regarding this issue—the Initiating Dialysis Early and Late (IDEAL) study, published in The New England Journal of Medicine in 2010 (2)—found that planned, early initiation of dialysis did not improve survival or other outcomes. However, IDEAL compared only two strategies, which achieved eGFR separation of just 1.8 (9.0 vs. 7.2) mL/min/1.73 m2. “That a kidney function outside this range exists at which starting dialysis is associated with better outcomes therefore remains possible, and uncertainty on this question among providers persists,” the researchers write.

Many previous observational studies have explored the optimal GFR to initiate dialysis, if any such threshold exists. In contrast to the trial, most of these observational studies found a strong survival advantage for late dialysis initiation. Why did the observational studies and IDEAL trial give such discordant results? In a close reading of the observational studies, Fu and colleagues found that virtually all had design errors leading to three types of bias, on top of residual confounding: immortal time bias, lead time bias, and collider stratification bias. (Fu explains this in detail in a recent Twitter thread (3).)

“These biases occur if investigators do not properly emulate the design of a clinical trial, in which the start of follow-up always aligns with the assignment of the treatment strategies,” Fu comments. “Fortunately, all three biases are self-inflicted and can be prevented by aligning start of follow-up and assignment of strategies.” The researchers used novel analytical methods—incorporating data cloning, censoring, and weighting—to “explicitly mimic” a multi-arm clinical trial comparing various dialysis initiation strategies.

The analysis included data on 10,290 patients with grades 4 to 5 CKD receiving routine nephrologist care between 2007 and 2017, drawn from the National Swedish Renal Registry. Median age was 73 years, 36% of patients were women, and 42% had diabetes. At baseline, 69% of patients had an eGFR between 15 and 20 mL/min/1.73 m2, with a median of 16.8 mL/min/1.73 m2. During follow-up, 3822 patients initiated dialysis. At a median follow-up of 3 years, 40.4% had died, and 23.8% had experienced a major cardiovascular event.

In their main analysis, Fu and colleagues compared outcomes for 15 dialysis initiation strategies based on eGFR values ranging from 4 to 19 mL/min/1.73 m2. In addition, in a secondary analysis, the authors investigated the same treatment strategies as the IDEAL study, to benchmark their results against the trial findings: early dialysis initiation was defined as an eGFR of 10–14 mL/min/1.73 m2 and late initiation as 5–7 mL/min/1.73 m2. The researchers also defined an “intermediate initiation” arm with an eGFR range of 7–10 mL/min/1.73 m2, representing the mean achieved eGFR in patients assigned to early initiation in IDEAL.

Five-year all-cause mortality and major adverse cardiovascular events (MACE; comprising cardiovascular death, nonfatal myocardial infarction, and nonfatal stroke) were compared between groups. The eGFR reference range was 6–7 mL/min/1.73 m2, the range at which most patients in Sweden start dialysis.

For all-cause mortality, outcomes were best for patients receiving very early dialysis, initiated at an eGFR of 15–16 mL/min/1.73 m2. In this category, 5-year absolute risk of death from any cause was 48.7% (95% confidence interval 43.9%–53.4%) compared with 53.8% in the reference range of 6–7 mL/min/1.73 m2. Absolute risk differences ranged from a 0.8% decrease at an eGFR of 5–6 mL/min/1.73 m2 to a 5.1% increase at 15–16 mL/min/1.73 m2. Associated hazard ratios were 1.01 and 0.89, respectively. Early initiation reduced mortality across patient subgroups defined by age, sex, diabetes, eGFR, and ischemic heart disease.

“Compared with starting at an eGFR between 6 and 7 mL/min/1.73 m2, we estimated that patients initiating at an eGFR between 15 and 16 mL/min/1.73 m2 would live on average 1.6 months longer over a 5-year follow-up period,” Fu stated.

However, to attain those extra weeks of survival, patients would need to start dialysis much earlier: 4 years earlier, on average. “For many patients, the modest survival benefit may not outweigh this increased time on dialysis,” the researchers write.

Absolute risk of MACE was lowest for patients initiating dialysis at an eGFR between 17–18 and 11–12 mL/min/1.73 m2, with progressively higher risks at later initiation. Compared with the reference range, absolute risk differences ranged from an increase of 1.5% to a decrease of 3.3%, with hazard ratios of 1.04 to 0.91, respectively. For earlier initiation at an eGFR between 15 and 16 mL/min/1.73 m2, absolute MACE risk was 2.9% (0.2%–5.5%) lower with a hazard ratio of 0.94 (0.91–0.98).

In a supporting analysis, following the GFR cutoffs used in the IDEAL study, early initiation at an eGFR of 10–14 mL/min/1.73 m2 was associated with a 3.3% (1.3%–5.3%) reduction in 5-year mortality and a 3.6% (1.0%–6.0%) reduction in MACE: hazard ratio 0.96 for both. Those results were “congruent” with the IDEAL findings, the researchers note, which found a hazard ratio of 1.04.

Rather than supporting a strategy of early initiation, Fu and colleagues believe that the modest survival benefit may not outweigh the substantially longer period spent on dialysis. The investigators conclude: “[T]hese data provide no support for any strategy other than starting dialysis on the basis of symptoms and patients’ preferences, which is widespread clinical practice, recommended by guidelines, and a [patient-centered] approach.”

References

  • 1.

    Fu EL, et al. Timing of dialysis initiation to reduce mortality and cardiovascular events in advanced chronic kidney disease: Nationwide cohort study. BMJ 2021; 375:e066306. doi: 10.1136/bmj-2021-066306

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 2.

    Cooper BA, et al. A randomized, controlled trial of early versus late initiation of dialysis. N Engl J Med 2010; 363:609619. doi: 10.1056/NEJMoa1000552

  • 3.

    Fu EL. @FuEdouard: “Why did observational studies find increased survival for late dialysis initiation, whereas the IDEAL RCT found no difference?Twitter, 6:30 p.m., Nov. 30, 2021. https://twitter.com/FuEdouard/status/1465840754820059146

    • PubMed
    • Search Google Scholar
    • Export Citation
Save