Design: systematic evidence map of clinical trials and observational studies
Population: 1,116 included studies of novel kidney biomarkers in critically ill patients, from 6,805 screened
Primary outcome: distribution of biomarker studies by research question
Effect: 944 prediction, 647 prognostication, 109 aetiology, 12 management, 6 trial enrichment

Novel biomarkers of acute kidney injury have been studied for two decades without settling into routine practice, and this systematic evidence map explains why. From 6,805 records screened, 1,116 studies of novel kidney biomarkers in critically ill patients were included and mapped by what question each asked.

The distribution is the argument. Of those studies, 944 addressed prediction of acute kidney injury and 647 prognostication of clinical outcomes. Only 109 addressed diagnosing the aetiology, 6 used biomarkers to enrich a clinical trial, and 12 studied biomarker-guided management. Adults made up 78.6% of studies and 93.3% used a cohort design; mixed critical care populations, cardiac surgery and sepsis accounted for 69.5% of the clinical contexts.

What that means at the bedside is that a biomarker result tells you something true and does not tell you what to do. The authors identify where implementation guidance does exist — surveillance of patients at risk, specifically those undergoing surgery and those exposed to multiple nephrotoxic drugs. Outside those settings, ordering a novel kidney biomarker will generate a number that no trial has connected to an action, and the honest position is to say so rather than to let the result drive escalation by implication.

Before ordering a novel kidney biomarker, decide what a positive result would change
The strongest implementation evidence is for surveillance in surgical and nephrotoxin-exposed patients
Count nephrotoxic drugs explicitly on the chart — multiple exposure is the identified risk context
Do not let a predictive result substitute for a management decision that has not been tested
Note that paediatric evidence is thin: fewer than a quarter of these studies included children

Why it matters

The literature is overwhelmingly about predicting kidney injury and almost silent on whether acting earlier helps.

The statistics, in plain English

Counts of studies measure research effort, not evidence quality — 944 prediction studies do not make prediction 78 times better established than management, they show where the field has spent its attention. The 93.3% cohort design figure matters for the same reason: cohort studies establish that a marker tracks with an outcome, while showing that acting on the marker improves anything requires the randomised management trials that number 12.