LAUSR

We thank Vodovar et al for their interest, and we agree there are difficulties in deciding the best way to apply Extracorporeal Treatments In Poisoning (EXTRIP) criteria. However, they have performed re-interpretation of our data which uses flawed methods and has reached a conclusion that could potentially mislead clinicians. The purpose of prediction in lithium poisoning is to identify early those patients who have a risk of developing severe consequences. Waiting 24 hours to make a prediction or using single cut-points (>2.5 mmol/L; Cr > 200 μmol/L or Li > 5.2 mmol/L), as they suggest are both sub-optimal strategies. We demonstrated for acute on chronic poisoning, extreme variability of the kinetics, the rarity of severe consequences and the absence of long-term sequelae. So whatever analysis they have done that concludes prediction with modelling or the nomogram is better in acute on chronic poisoning (done without access to the data) is incorrect. In contrast, chronic lithium poisonings have highly predictable kinetics, where we demonstrated future concentrations can be modelled to within ±0.3 mmol/L from simply the estimated creatinine clearance and the lithium concentration on admission. It is not useful to use sensitivity or specificity to calculate the percentage of patients that will have lithium >1 mmol/L at 36 h, as if an error of 0.3 mmol/L was important. The key question is whether the nomogram included all patients who would develop neurological sequelae (while estimating the serum lithium concentration >1 mmol/L at 36 h post admission), and who should therefore be considered for ECTR. Our nomogram accurately predicted all patients who developed neurological sequalae, including the few patients who received ECTR, and all but two patients who would have a lithium concentration >1.4 mmol/L at 36 h post admission. The only role of the nomogram is in chronic lithium toxic patients. Although we accept further validation of the accuracy of prediction those with sequelae would be useful, identifying that a proportion of people will be slightly over or above their predicted concentration is not. Finally, their interpretation of EXTRIP and the Paris criteria appear to endorse a practice of allowing very severe chronic lithium toxicity to languish for 24 h before a decision is made to do ECTR. Further, they have implausible assumptions about the half-life of lithium based on normal individuals, and no evidence base behind these assumptions. The Paris criteria would have identified only two of the six individuals in our study who developed sequela and appeared also to miss nine out of 24 patients who developed sequelae in their series also. We believe our adaptation of the EXTRIP guidelines provides a much better means to translate the work of the EXTRIP systematic review into practice.