Conflicts of Interest
None to report
We thank Cong and colleagues for their interest in reading our study and
the comments they have brought to our attention. It is important to note
that the study was focused on a population of non-emergent patients. We
do agree that prior risk scores for predicting cardiac surgery
associated acute kidney injury (CSA-AKI) all have merit, but our study
is unique in the population being assessed (1, 2). We also agree that
mild AKI is a significant complication and is likely predictive of poor
post-operative outcomes, hence we did present data regarding the outcome
of any AKI, and it is important to appreciate that the scorecard was
derived using the outcome of any AKI. We primarily provided an example
of application of the scorecard to generate a risk prediction for severe
AKI acknowledging that it may have even more immediate clinical
importance due to the subsequent morbidity, mortality and healthcare
burden.
With regards to the scorecard, both B-coefficients and odds ratios (ORs)
are interchangeable in this situation. Each integer of the score was
based on 0.5 increments of the B-coefficients. In addition, we have
added the B-coefficients scores in the supplementary materials for
reference. We acknowledge that age, male sex and complexity of procedure
while associated with AKI were not statistically significantly
associated with severe AKI. We suspect this relates to the smaller
numbers of outcomes for severe AKI and that had there been a similar
number of events, the results may have more closely aligned with any
AKI. Ultimately, it was the outcome of any AKI that was used to derive
the model, but the provided example scorecard was that of severe AKI.
We do agree that there is value in using intraoperative and
postoperative variables to predict CSA-AKI. We certainly acknowledge
that the use of only preoperative variables may have limited
discrimination for post-operative events. This emphasizes the need for
validation of our risk score in a different CSA population. We still do
feel that there is value in preoperative scorecards to help facilitate
shared decision making about surgery itself, most importantly for this
sub-population of individuals who are non-emergent. Future work that is
planned will be to evaluate and compare the performance of this
predictive model to existing risk scores and dynamic models as
highlighted, most importantly in this non-emergent population.