Understanding probability and exposure in paracetamol overdose risk assessment

To the Editor:

We thank Hossein Sanaei-Zadeh for his comments. A number of previous studies have investigated whether dose can predict outcomes, but they do not provide a continuous probability of requiring N-acetylcysteine (NAC) versus the dose. For example, the study by Waring et al. used a single cut-off which confirms dose being a predictor but cannot guide treatment over a range of doses.¹ This study cannot answer the question, “Do we treat with NAC at 13 g or not treat at 13 g?” Thomas et al. and Zyoud et al. reported smaller studies of 300–400 patients with similar single threshold cut-offs.^2,3 The aim in our study was to provide a probability of requiring NAC over a large dose range to assist clinicians in making early treatment decisions.⁴ However, we do not provide a prescriptive approach to the treatment of paracetamol overdose, and in particular we do not give a definite “cut-off dose” for a “wait and see approach”. We hope the study will assist clinicians in making individual treatment decisions. A prospective study will be required to determine if using a general cut-off dose for all cases for early treatment decisions, is feasible and safe.

Dr Sanaei-Zadeh indicates that the apparent effect of single dose-activated charcoal (SDAC) appears to reduce with increasing dose and indeed becomes almost insignificant above 60 g doses. We would like to highlight that the nature of logistic regression means that the greatest influence of any intervention will be apparent at a probability of the event being closest to 0.5. Hence, the closer the probability of an event gets to either 0 or 1 the more condensed the effect appears to be on the probability scale. However, the odds ratio (OR) of the influence of the effect is unaffected by the underlying probability of the event. If a patient took an overdose of 60 g then although SDAC would reduce the OR of being above the treatment line (which will be the same for all overdose levels) the apparent probability that SDAC will reduce the exposure sufficiently to be below the treatment line will be diminished, because the underling probability of being above the “150” line is almost one. It is important to note that this does not mean that SDAC will not reduce exposure and hence even though the probability of being above the line is only marginally affected the total exposure will be reduced. It therefore depends on why SDAC is being administered? If it is to reduce the probability of needing NAC, then it will be futile at very high doses. If it is to reduce the overall exposure to paracetamol and provide some potential improvement of clinical outcomes, then it would seem very appropriate. For example, administering SDAC in a patient taking 60 g could potentially prevent cardiovascular and metabolic complications known to occur with massive paracetamol overdoses.

The study did not assess whether vomiting affected the paracetamol concentration because the timing of vomiting was not well recorded, which is essential in such an analysis. Early spontaneous emesis may reduce the fraction absorbed but delayed vomiting due to paracetamol toxicity or NAC is unlikely to affect drug absorption.

Finally, we hope that this study provides clinicians with additional information for them to make an informed decision when treating paracetamol overdose, whether in a location where paracetamol concentrations are not routinely available, or in hospitals where paracetamol concentration is a standard part of the risk assessment.