Automated alerts for acute kidney injury warrant cautionBMJ 2015; 350 doi: https://doi.org/10.1136/bmj.h19 (Published 07 January 2015) Cite this as: BMJ 2015;350:h19
All rapid responses
Sawhney makes the important point that the number of AKI alerts will be determined by how the baseline is defined and the level of change required to exceed a somewhat arbitrary cut off. He highlights the difficulty of defining the baseline and how this may lead to large differences in reporting.
Another important point is the misconception that AKI is something to do with the kidneys themselves in the vast majority of cases. With the exception of the small percentage of cases with primary kidney disease, AKI is essentially a marker of severity of illness. The rise in creatinine could be considered as analogous to the (now) ubiquitously measured C-reactive protein, an entirely non-specific marker of illness with a different, but overlapping, range of causes. Any rise in creatinine means the patient is ill.
The main advantage of the alert system is that is should draw somebody's attention to the fact that the patient is ill AND lead them to address any reversible factors and treat the underlying cause. Only the early identification of the small number of cases with primary kidney disease or kidney failure sufficiently severe to warrant renal replacement therapy that should trouble Dr Sawhney's on call renal pager.
Competing interests: No competing interests
We read Dr Sawhney’s commentary on automated detection systems for AKI with interest. We agree with several of the points made and would like to add some additional clarity around the NHS England Patient Safety Alert (NHS/PSA/D/2014/010).
The last five years has seen a burgeoning interest in electronic systems to detect AKI (based on a change in serum creatinine concentration) and then to notify or alert relevant clinicians. The clinical need that such systems are attempting to address is clear. Delays or failures to recognise AKI lead to deficiencies in patient care across all specialities , and there are several studies reporting an improvement in physician response time when AKI alerting systems are introduced [2, 3]. However, we agree that a diagnosis of AKI is a clinical one, not just to consider the performance of AKI detection in the context of the individual patient (as one should for any biochemical test) but to determine the underlying cause - it is vital that AKI is not regarded as a single entity, rather a syndrome with multiple differing aetiologies. These principles are reflected in both the name of the new AKI test (AKI Warning Stage) and its suggested accompanying text comment: “Rise in creatinine may indicate Acute Kidney Injury stage (1, 2 or 3 as appropriate), please review urgently” . These guidelines have been circulated to all laboratories in England and are available on the NHS England AKI Programme’s “Think Kidneys” website. This approach therefore mirrors that described by Dr Sawhney.
The internationally accepted criteria for defining AKI are in fact very precise. The difficulties to which Dr Sawhney refers arise with their use in clinical practice, when there is an absence of prior serum creatinine measurement within the last seven days with which to compare a current value. In this situation, there have been a variety of different proposals and of course different definitions of baseline creatinine will change the performance of AKI diagnostic criteria, as previously described and as Dr Sawhney’s as yet unpublished data show [5-7]. Using the lowest serum creatinine concentration from the last twelve months results in significant over-diagnosis when compared to the ‘gold standard’ of the human eye (a panel of nephrologists looking at previous results supplied in a way analogous to usual clinical practice); an approach using an average from the same time period has better agreement . In part, the NHS England Patient Safety Alert was issued to bring standardisation to an increasingly varied application of AKI detection in clinical practice, and the recommended algorithm uses a dual approach to the choice of baseline serum creatinine. The lowest previous creatinine result is used as baseline if measured within the previous 7 days and if not present, the algorithm uses the median of all previous results within the last year .
Throughout the NHS England AKI Programme there is a strong focus on measurement . This includes utilisation of prospective data collection from the AKI detection algorithm, to compare variants of AKI detection algorithms, to link algorithm performance to outcomes and to begin to compile such data nationally. In this way, ongoing evidence generation can occur alongside quality improvement efforts and will direct future development of AKI detection.
Robert Hill Consultant Clinical Biochemist and Chair Detection Workstream, NHS England “Think Kidneys” AKI Programme, UK Renal Registry, Bristol BS10 5NB. Robert.Hill1@nhs.net
Nicholas Selby Consultant Nephrologist and Co-Chair Detection Workstream. NHS England “Think Kidneys” AKI Programme, Department of Renal Medicine, Royal Derby Hospital, DE2 3NE.
Competing interests: No competing interests