Causality assessment tool for drug‐induced acute kidney injury in patients admitted to intensive care units: Requirements from a multicentre retrospective study in the Netherlands reusing data from electronic health records

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Aim This study aimed to provide key requirements for a drug‐induced acute kidney injury (d‐AKI) causality assessment tool (CAT) for intensive care unit (ICU) patients, using vancomycin‐induced AKI (v‐AKI) as a test case. Methods A retrospective review was conducted by medical experts of electronic health records of ICU patients from 14 Dutch ICUs. At each ICU, a panel consisting of three medical experts was established. Each case was presented via an electronic case report form (eCRF) including criteria from the Naranjo and World Health Organization‐Uppsala Monitoring Centre CATs, both enhanced by d‐AKI features (N‐CAT‐d and W‐CAT‐d, respectively). After completing the eCRF, two causality scores based on the two CATs were presented to the experts, who subsequently provided their final judgement. Inter‐rater variability was assessed using percentage agreement. The final panel judgement was established by consensus. Results In total, 41 ICU patients were assessed, resulting in 123 individual and 41 final expert panel judgements. The overall mean percentage agreement for the causality score was 34.4%, and that for the d‐AKI‐specific alternative aetiologies criterion was 70.0% (W‐CAT‐d) and 78.6% (N‐CAT‐d). The final panel judgements resulted in 15 cases (36.6%) being deemed possible or probable v‐AKI and aligned with W‐CAT‐d causality scores in 24 (58.5%), aligned with N‐CAT‐d scores in 12 (29.3%) and diverged from both in 14 (34.1%) cases. Conclusions The W‐CAT‐d showed greater alignment with the clinical judgements of ICU medical experts. The d‐AKI‐specific alternative aetiologies feature supports consistency in the evaluation of this crucial criterion. Additional refinements are necessary to better capture the complexity of ICU d‐AKI cases.