Time’s Up! Start Dialysis Later in Children

In adults, there had been a consistent trend toward initiating dialysis later, at lower levels of kidney function (eGFR).¹ Observational studies have had conflicting results, with more recent studies indicating that early dialysis initiation may be harmful.^2,3 Only a single trial informs clinical practice: the Initiating Dialysis Early and Late (IDEAL) trial, published in 2010. This trial randomized 828 adults with progressive CKD to start dialysis at an eGFR of 10–15 or 5–7 ml/min per 1.73 m².⁴ Although the separation between groups was lower than intended at 2.2 ml/min per 1.73 m², there was a 6-month difference in the timing of dialysis initiation between the two groups. Uremic symptoms were cited as the reason for protocol violation among most patients in the late-start group. The study, which included a wide range of outcomes, concluded there was no improvement in mortality or quality of life, and that costs were higher in the early-start arm. Although potentially underpowered, and with less between-group separation than was planned, IDEAL demonstrated that starting dialysis based on eGFR alone is unlikely to benefit patients. What is indisputable is that starting dialysis “early,” means “early” exposure to the physical and psychosocial complications of dialysis. The publication of IDEAL has coincided or lead to patients starting dialysis later, with the average eGFR at initiation of dialysis among adults in the United States falling since 2010.²

Few data exist to guide decision making regarding the timing of dialysis initiation in children. It is also likely that the results of the IDEAL trial are not transferable to children. The spectrum of underlying kidney diseases differs, and with it the trajectory to RRT. The prevalence of comorbidities is lower, particularly cardiovascular disease, prognosis is more favorable, and the nature of dialysis and its impacts on children and their families is very different. In this context, the study of Winnicki et al. published in this issue of the Journal is itself timely.

The authors examined temporal trends in the timing of dialysis initiation among children and associations between the timing of dialysis initiation and mortality, using United States Renal Data System (USRDS) registry records for 15,170 children (1–18 years) starting dialysis between 1995 and 2015.⁵ Early dialysis initiation, defined at an eGFR>10 ml/min per 1.73 m², occurred in 29.0% of children overall and increased markedly over time from 16.6% in 1995 to 40.7% in 2015. The median age was similar between the two groups, but with substantial negative skewness, which was differential, there were more young children (1–5 years old) among the early-start group. Those starting dialysis early were also more likely to be white, female, have GN as a cause of ESKD, and be either underweight or obese. The median length of follow-up was 6.1 years for the early-start group and 9.6 years for the late-start group. Patients that started dialysis early had a higher mortality rate than those who started later, with the unadjusted 1-year survival rate among the early-start group at 95.6% compared with 97.7% among the late-start group.

Cardiovascular events were the most common cause of death. Despite adjustment for covariates, including those noted to be unevenly distributed between groups, the mortality difference persisted. The authors tested for and found an interaction between dialysis modality and outcome, such that the increase in risk associated with early dialysis initiation was more pronounced among children receiving hemodialysis (hazard ratio, 1.56; 95% confidence interval, 1.29 to 1.75). The effect size was smaller and no longer statistically significant among children receiving peritoneal dialysis (PD) (hazard ratio, 1.07; 95% confidence interval, 0.91 to 1.25). The authors suggested that a superior preservation of residual renal function with PD compared with hemodialysis may have accounted for this observation. That PD better preserves renal function, at least initially, is well established and may be relatively more important in children, who are more likely to have meaningful urine output at dialysis initiation due to tubulointerstitial causes of ESKD being more common.⁶ The interaction between dialysis modality and outcomes observed in this study needs to be considered when translating the results to settings where HD or PD are the predominant modality.

The risks of bias inherent to observational studies are well known. There are a number of tools for evaluating risk of bias in this study design, most notably the Newcastle Ottawa Scale and the Cochrane non-randomized studies tool.⁷ Chief among the potential biases, is a form of selection bias, indication bias, where participants are differentially chosen to receive an intervention because of predicted differences in prognosis.⁷ Such bias results in the intervention appearing to lead to poorer outcomes—sicker patients typically receive more aggressive treatment.⁸ Indication bias is notoriously difficult to adjust for and the direction of bias is usually toward worse outcomes among the treatment group.^8,9 As a result, this study is likely to have over-estimated the negative impact of early dialysis initiation. There is also the potential for survivor bias, whereby children with a later dialysis start date are a healthier cohort because they have already demonstrated survival to that point. Another important limitation of this study is an inability to consider important factors specific to children, such as the impact of dialysis initiation on growth and development, given the limitations of the USRDS dataset.

That the proportion of children starting dialysis with an eGFR>10 ml/min per 1.73 m² has more than doubled in the last 20 years is concerning given the absence of any benefit found in this study, and in the adult studies, including the IDEAL trial.¹⁰ There appears to be no trade-off. The direct, immediate and incontrovertible deleterious financial, psychosocial, and physical impacts of dialysis are experienced on a daily basis by clinicians, children, and their families.¹¹ A recently published, independent analysis also using USRDS data reached similar conclusions to the Winnicki study.¹⁰ Although more research is clearly needed to address this question, including prospective cohort studies with information on pediatric specific outcomes such as growth and development, until then the implications are clear. Time’s up. Start dialysis later, except in the context of a trial.

Disclosures

Dr. Craig was an investigator and author for the IDEAL study. Dr. Larkins has nothing to disclose.