Incidence and Adverse Outcomes of Acute Kidney Disease: A Systematic Review and Meta-Analysis.

Abstract

Rationale & objective: Estimates of the incidence of acute kidney disease (AKD) and its associated adverse outcomes are inconsistent, which may be due, in part, to differences in prior studies' definitions of AKD. This study sought to summarize these reports and identify study-level characteristics, including the definition of AKD, that may explain the observed heterogeneity.

Study design: Systematic review and meta-analysis.

Setting & study populations: Adults aged ≥18 years and not receiving maintenance kidney replacement therapy.

Selection criteria for studies: Observational studies that assessed the incidence of AKD and its association with adverse outcomes.

Data extraction: Two reviewers independently extracted data and assessed study quality.

Analytical approach: AKD definitions were classified as: 1) Acute Disease Quality Initiative (ADQI) or ADQI-equivalent or 2) KDIGO/KDIGO-equivalent. Random effects meta-analysis was used to calculate pooled estimates of incidence and the relationship between AKD and outcomes([mortality, kidney failure, onset of chronic kidney disease [CKD]) summarized by odds ratios (ORs) and 95% confidence intervals (CIs).

Results: Among 1,883 identified studies, 59, involving nearly 6 million participants, met inclusion criteria. Most studies were classified as having good quality per the New-Castle Ottawa scale (n=44). The pooled incidence of AKD was higher when defined by ADQI/ADQI-equivalent compared with KDIGO/KDIGO-equivalent criteria (26.6% [95% CI, 20.3-34.9%] vs. 11.1% [95% CI, 7.6-16.3%], p for difference <0.001). The pooled odds ratio of all-cause mortality associated with AKD was similar whether defined with KDIGO/KDIGO equivalent or ADQI/ADQI-equivalent criteria (3.8 [95% CI, 2.2-6.7] vs. 3.0 [95% CI, 2.1-4.4], p =0.5). After accounting for the baseline AKI status, both the incidence of AKD and its association with all-cause mortality were similar for the two definitions. The incidence of AKD was 13.6% and 11.1%, and the ORs for all-cause mortality were not different (4.2 [95% CI, 2.0-8.7] vs. 3.8 [95% CI, 2.2-6.7], p =0.8) using the ADQI/ADQI-equivalent and KDIGO/KDIGO-equivalent definitions, respectively. Similar results were observed for the association between AKD and development of CKD, but the association between AKD and kidney failure was stronger in studies using the KDIGO/KDIGO-equivalent definition.

Limitations: Heterogeneity persisted across most of the examined subgroups.

Conclusions: Estimates for AKD incidence and AKD-associated risk for clinical outcomes vary by the definition used for AKD. These findings inform the assessment of the incidence and consequences of AKD in research and clinical settings.