DISCUSSION
In this single center retrospective analysis, in whom hemodialysis was performed or recommended by our PCC, when the EXTRIP and Paris criteria are both in favor of hemodialysis, dialyzed patients generally had favorable outcomes. When the two criteria are against hemodialysis, non-dialyzed patients also had favorable outcomes. As reported by its creators, application of the Paris guidelines prompts dialysis in fewer patients overall, but when the criteria were discordant, the EXTRIP criteria outperformed the Paris criteria at identifying potentially ill patients who might benefit from hemodialysis. Of the cases of deterioration not identified by the Paris criteria, two were not identified by the EXTRIP “recommends” criteria, and all four were identified by the EXTRIP “suggests” criteria. EXTRIP criteria were not intended to be merged for comparative analysis, but this comparison is presented here to reflect the merged EXTRIP analysis performed in the original Paris validation study.
Descriptive data are unfortunately limited for these cases. We cannot completely exclude processes unrelated to their lithium toxicity contributing to patients’ deterioration. This qualification, however, is equally applicable to all cohorts and reflects the typical quality of data from poison control center databases. Furthermore, poison center data can be limited by completeness and accuracy for coded clinical effects and treatments.[10],[11] There may have been variability between original PCC coders for clinical effects such as “confusion,” “altered mental status,” and “other-miscellaneous.” Narrative accounts can be selective in their inclusion/exclusion of relevant data and can terminate in many cases before the outcome of interest is determined in a subsequent retrospective study. Long term outcomes such as SILENT cannot be extrapolated from poison center cases, for example, as routine follow up on the order of months or years is seldom performed.
Additionally, our cohort possibly suffers from two sources of selection bias. There may have been cases in which EXTRIP or Paris criteria were met, but dialysis was neither recommended nor performed. These cases would not have been identified by our search criteria but were unlikely to be significantly ill since most sources prior to EXTRIP suggest more liberal indications for hemodialysis. Additionally, we only reviewed cases reported to one regional metropolitan poison center in the United States. Since poison center reporting is largely voluntary, we may have missed cases in our catchment area not reported to our center.[12]