A technical correction to an influential kidney-disease analysis
Researchers have updated a trial-level analysis examining whether changes in glomerular filtration rate, or GFR, slope can serve as a surrogate endpoint for kidney failure in clinical trials. The correction appears in an addendum to a Nature Medicine article first published in June 2023.
The update concerns the treatment of shared control groups in multi-arm trials. Its authors report that the original analysis incorrectly stipulated the shared control groups in the African American Study of Kidney Disease, known as AASK. They repeated the meta-regression using the corrected designation.
The revised summary results differed from the published results only in the hundredths place, according to the addendum. The authors have supplied updated trial-level association summaries in a new figure and table.
Why the correction matters
The original work used a Bayesian meta-regression across randomized controlled trials to quantify the association between treatment effects on GFR slope and treatment effects on clinical endpoints. Such an analysis brings together evidence from separate trials to examine whether a measurement that changes earlier can reliably track a later clinical outcome.
In this case, the clinical endpoint included kidney failure requiring replacement therapy, GFR below 15 mL/min per 1.73 m2, or doubling of serum creatinine. The trial-level analysis compared treatment effects on those outcomes with treatment effects on the GFR slope.
Multi-arm trials require particular care when they include a control group shared by more than one intervention arm. Comparisons that use that common control are not statistically independent in the same way as entirely separate trials. The authors state that their approach accounted for correlated sampling errors created by multi-arm studies with shared controls.
The addendum says the error lay in how the AASK shared-control groups were stipulated, rather than in the overall purpose of the analysis. Repeating the work with the correct trial designation changed summary results slightly, but did not produce a large numerical revision in the authors’ account.
What the revised presentation shows
The updated figure presents separate studies as circles, with circle size proportional to the number of clinical events. The events counted are kidney failure with replacement therapy, GFR below 15 mL/min per 1.73 m2, or doubling of serum creatinine.
Circle colors identify intervention types. The figure includes a regression line through the studies and 95% pointwise Bayesian confidence bands. Triangles identify studies whose estimated treatment effects extend beyond the plotted margins.
The intervention labels listed in the addendum include renin angiotensin system blockers, calcium channel blockers, blood-pressure interventions, DPP-4 therapies, GLP-1 therapies and SGLT2 therapies. The updated table provides the associated trial-level summaries.
A narrow revision with practical value
Addenda can be easy to overlook because they often concern technical details rather than a new therapy or a new clinical trial. But the details matter when a study addresses how trials measure progress toward severe disease outcomes.
The authors’ correction is also a reminder that evidence synthesis depends on decisions about trial structure as well as on the data reported by individual studies. A shared control group is not merely a formatting detail in a dataset: it affects the relationship among estimates that enter a pooled analysis.
For readers of the original meta-analysis, the immediate message is measured. The authors report that the corrected results differ only at the hundredths place. Still, they have replaced the affected trial-level summaries so that the published record reflects the corrected specification.
The addendum does not describe a new intervention result or a new patient dataset. Instead, it updates the statistical handling of an existing body of randomized-trial evidence. That distinction is important: the revised material is a correction to the quantitative summary of the association under study.
What comes next
The corrected figure and table now sit alongside the original Nature Medicine publication. Researchers and clinicians using the article’s trial-level estimates can refer to the addendum for the updated summaries.
The episode illustrates a broader feature of clinical research: published analyses can be refined when an issue is identified, even after a study has become widely accessed. The authors report more than 4,000 accesses for the article record and have made the correction explicit rather than leaving the prior specification unaddressed.
For the kidney field, the central question remains the one the original analysis set out to quantify: how closely treatment-related changes in GFR slope correspond with treatment-related effects on major clinical kidney endpoints across randomized trials. The new addendum updates the numbers used to describe that relationship while reporting only very small differences from the original summary results.
This article is based on reporting by Nature Medicine. Read the original article.
Originally published on nature.com






