Early and accurate acute kidney injury (AKI) detection may improve patient outcomes and reduce health service costs. This study evaluates the diagnostic accuracy and cost-effectiveness of NephroCheck and NGAL (urine and plasma) biomarker tests used alongside standard care, compared with standard care to detect AKI in hospitalised UK adults.
A 90-day decision tree and lifetime Markov cohort model predicted costs, quality adjusted life years (QALYs) and incremental cost-effectiveness ratios (ICERs) from a UK NHS perspective. Test accuracy was informed by a meta-analysis of diagnostic accuracy studies. Clinical trial and observational data informed the link between AKI and health outcomes, health state probabilities, costs and utilities. Value of information (VOI) analysis informed future research priorities.
Under base case assumptions, the biomarker tests were not cost-effective with ICERs of £105,965 (NephroCheck), £539,041 (NGAL urine BioPorto), £633,846 (NGAL plasma BioPorto) and £725,061 (NGAL urine ARCHITECT) per QALY gained compared to standard care. Results were uncertain, due to limited trial data, with probabilities of cost-effectiveness at £20,000 per QALY ranging from 0 to 99% and 0 to 56% for NephroCheck and NGAL tests respectively. The expected value of perfect information (EVPI) was £66 M, which demonstrated that additional research to resolve decision uncertainty is worthwhile.
Current evidence is inadequate to support the cost-effectiveness of general use of biomarker tests. Future research evaluating the clinical and cost-effectiveness of test guided implementation of protective care bundles is necessary. Improving the evidence base around the impact of tests on AKI staging, and of AKI staging on clinical outcomes would have the greatest impact on reducing decision uncertainty.
We are grateful to Thomas Walker and Rebecca Albrow at NICE for their thoughtful comments on earlier versions of the economic model and to the NICE Diagnostic Committee for their critical review of our identifed evidence. We are also grateful for the advice and clinical guidance received from the NICE Specialist Advisory Group for DG19 and to Peter S Hall and Alison F Smith (on behalf of the team) for providing early versions of their economic model that was instrumental in the development and structuring of the model used in this study. A big thank goes also to Lara Kemp for her secretarial support and patience throughout the study. The results presented in this paper have not been published previously in any academic journals, nor have they been submitted elsewhere. This work has informed the development of NICE guidance for diagnostic testing for AKI (https://www.nice.org.uk/guidance/dg39) and a full report to the funder describing the totality of this work will be published in the NIHR, HTA mono‑ graph series in due course.
The fndings presented in this manuscript are part of a broader research project funded by the National Institute for Health Research (NIHR) and com‑missioned through the NICE Diagnostic Assessment Programme (project no 12/88/97). The views expressed are those of the authors and not necessarily those of NICE, the NHS, the NIHR or the Department of Health. The Health Economics Research Unit and the Health Services Research Unit, University of Aberdeen, are funded by the Chief Scientist Ofce of the Scottish Government Health and Social Care Directorates.
Data Availability StatementNo new individual patient data were generated in support of this research. All cost‐effectiveness and valuation of information model parameters are reported within the article and additional supplementary materials.
- Acute kidney injury
- Critical care
- diagnostic accuracy
- Economic evaluation
- Markov model