When is research clinical advice? Interpreting an exploratory study of paracetamol overdose

To the Editor:

We thank Drs. Rahman, Gilberti and Berman for their comments although we are concerned that they interpret parts of the discussion as conclusions that should immediately change clinical practice. In fact, we conclude that the reported dose can be used as a predictor of patients who may require N-acetylcysteine (NAC), which can improve risk assessment. We do not provide any hard and fast recommendation, but discuss some scenarios where this analysis could be used to influence clinical practice. For example, the reported dose could potentially allow earlier administration of NAC in large overdoses where NAC is usually not commenced until a paracetamol concentration is available. Our suggestion of 50 g is an arbitrary value chosen to make a discussion, and 100 g, 60 g or 30 g could also have been chosen to illustrate the same point. The converse example is that some countries treat all patients ingesting paracetamol with NAC,¹ so this study could allow the more targeted use of NAC, rather than treating all patients. Finally, paracetamol concentrations are either not available or difficult to access in many resource-poor countries,² and this study could potentially improve the risk assessment where dose may be the only information available.

We agree that only 10% of the doses in this study were greater than 25 g (NB. 25 g carries an approximate 50% chance of requiring NAC). However in this data set, this corresponds to 148 (of 1571) dose events and hence we believe that doses greater than 25 g in numbers are reasonably well represented. While it is always true that the tails of any distribution will carry error, it should be noted that the logistic distribution, and hence its cumulative density (which is used as the structural form of logistic regression), is a symmetric function around the probability of 0.5. Therefore accurate characterisation of the function below probability 0.5 (doses < 25 g) will help inform the function at probability greater than 0.5 (doses > 25 g).

The intention of this study was to provide clinicians with more information to assist in making a risk assessment in paracetamol overdose. We also hope that it will encourage further research into improving the treatment of paracetamol poisoning.

Declaration of interest

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.