The authors reported a correction to their trial-level Bayesian meta-regression that examined how treatment effects on GFR slope relate to treatment effects on clinical endpoints of kidney failure. The analysis explicitly modelled the correlated sampling errors that arise in multi-arm randomized trials with shared control groups. The Addendum states that the original publication had mis-specified the shared-control structure for the African American Study of Kidney Disease (AASK). The study team re-ran the meta-regression with the correct multi-arm designations and found that the summary results changed only in the hundredths place relative to the originally published numbers.
The Addendum replaces the trial-level visual and tabular summaries. Figure 1 shows two bubble plots that compare study-level treatment effects on GFR slope with treatment effects on the clinical endpoint. The Addendum reports two fitted regression lines: the line for the 3-year total GFR slope had an R squared of 0.95 and a meta-regression slope of −0.35. The line for the chronic GFR slope had an R squared of 0.55 and a meta-regression slope of −0.33. Both fitted lines decline, indicating that larger treatment effects on the GFR slope correspond to lower hazard ratios for the clinical endpoint in these trial-level analyses. Blue bands around the lines depict 95% pointwise Bayesian confidence intervals.
Each study appears as a circle on the scatterplots. The circle size is proportional to the number of events, which the authors defined as kidney failure with replacement therapy, an estimated GFR below 15 mL/min per 1.73 m2, or a doubling of serum creatinine. Circles were coloured to indicate intervention class; the Addendum maintains the classification scheme used in the original analysis (for example, RASB, CCB, blood-pressure interventions, DPP-4 inhibitors, GLP-1 agents, SGLT2 inhibitors and other categories listed in the main article). Triangle symbols identify studies whose estimated treatment effects lie beyond the plotting margins.
The regression lines and their fitted slopes are derived from a Bayesian meta-regression framework that accounts for correlated sampling errors across arms sharing a control group. The Addendum specifies that correcting the AASK multi-arm designation and repeating the analysis produced negligible numerical differences, limited to hundredths, relative to the prior report.
The Addendum refers readers to Table 1 for trial-level analyses by subgroups. The table in the Addendum provides the updated numerical summaries that correspond to the plots in Figure 1. The re-analysed outputs reaffirm the relative tightness of the association for the 3-year total GFR slope (higher R2) versus the chronic slope (lower R2) in this trial-level comparison across 66 studies.
The Addendum lists the full author group and institutional affiliations, mirroring the multi-centre collaboration reported previously. The author list again identifies members of the CKD-EPI Clinical Trials Consortium and investigators from multiple academic nephrology centres and research units internationally. The Addendum also cites the original article (Nature Medicine, 2023) as the methodological source for the meta-analysis and directs readers seeking full trial-level inclusion criteria and analytic details to that publication.
The Addendum is explicit about the nature of the correction: a mis-specification of shared controls in one multi-arm trial (AASK) that was corrected and re-analysed. The authors note only minimal numeric change after correction (differences in the hundredths place). The Addendum supplies corrected Figure 1 and Table 1 but does not re-state detailed methodological parameters, such as exact priors or full modelling formulas; those elements remain documented in the original article cited in the Addendum. Readers who require the full modelling specification, the complete list of included trials, or subgroup analytic code are referred to the original 2023 publication for the comprehensive methods and supplementary materials.