The reflections of Hickman and colleagues [1] are challenging. The authors argue strongly for a translational approach within the addiction sciences, with the relationship between alcohol and opioid overdose being used as a prime example. At one level, it is difficult to dispute a call for different fields of research to interact in addressing major questions in the drug and alcohol field. On the other hand, it is not clear to me that laboratory research, and animal-based research in particular, necessarily provides more clarity on clinical and epidemiological issues. In fact, I would argue that in some cases, and overdose is one such case, such research may detract from our understanding of real-life behaviours and clinical issues. There is also a risk of reductionism inherent in such an approach, a fallacy that may most certainly lead us further from coherent explanation.

As stated above, laboratory research may, in fact, lead us astray rather than to the truth. The example of conditioned opioid tolerance in rat research is germane [2]. Broadly speaking, the argument from such research is that opioid tolerance is a conditioned response, and that tolerance is lowered in unfamiliar environments. Thus, the popular scenario of an overdose death in the street would be seen as an example of such reduced tolerance when faced with unfamiliar cues. The theory is popular among animal researchers and has generated much research. I am often asked how it translates to human overdose. I respond by stating that I believe it is of almost no clinical relevance at all, and is a distraction at best. Humans are not rats. The overwhelming majority of overdoses, fatal and non-fatal, occur in familiar home environments. Moreover, they occur among a population that has high levels of psychopathology (e.g. major depression, borderline personality disorder) and systemic disease [3]. The latter include serious hepatic, cardiac and pulmonary pathology. Animal models can never take into account such factors, and I believe provide no explanatory model for opioid overdose among the dependent drug user. They simply do not translate to the complexity of the human opioid user.

In fact, the example chosen by Hickman and colleagues, alcohol and overdose risk, is an odd one. It is not clear to me what the debate here actually is. While we may not understand clearly all the mechanisms or circumstances contributing to overdose, the role of alcohol and other central nervous system (CNS) depressants is one of which we are very confident. In fact, given everything we know about CNS depressants, what would have been extraordinary would be findings showing alcohol was not a factor. The epidemiological, toxicological and behavioural data all point in the same direction: using opioids in the presence of alcohol increases the risk of overdose. Furthermore, there are plausible pharmacological reasons to expect this. It is not as if this is something new, or that it is exclusive to opioids [4]. Any anaesthetist will advise against the use of alcohol prior to an anaesthetic.

I alluded to the issue of reductionism above. Essentially, this posits that explanations at a ‘more basic’ level are superior to those at the higher levels. Generally, this is taken to mean that physics, for example, provides a deeper explanation in some way than psychology. In the case we are dealing with here, explanations derived from pharmacological manipulations of lower order species are being argued to provide more fundamental knowledge of human overdose. The problem is that information is lost as we go ‘deeper’. For example, interactions between quarks do not provide explanations for the group behaviours of humans, even though we are made of quarks. It is only by studying humans themselves that we can understand such behaviours. Similarly, we lose enormous amounts of information and explanatory power in attempting to understand the deaths and near-deaths of opioid users if we look to the laboratory, rather to the events themselves.

To summarize, there is no doubt that Hickman and colleagues are correct in that we can learn from other fields of research. We need, however, to avoid the assumption that such ‘basic’ research is necessarily more informative than studies of actual people. They may, in fact, be a hindrance to our clinical understanding.