Extracting value from case reports: lessons from serotonin toxicity


Serotonin toxicity (ST) is an iatrogenic drug-induced ‘toxidrome’ displaying the characteristics expected of a synaptic serotonin concentration-related phenomenon. The term ST is preferable because serotonin syndrome (SS) implies an idiosyncratic reaction like malignant hyperthermia (MH) or neuroleptic malignant syndrome (NMS). Serotonin toxicity is important because fatal combinations of therapeutic drugs are sometimes inadvertently combined. Also, there is incomplete and erroneous information concerning which drugs are capable of precipitating ST, and its symptoms and treatment. Much of this misleading information is propagated through the publication of case reports, most of which have not included a systematic review and fail to report the key findings, positive and negative, that might allow a more precise judgement concerning the reliability and value of such reports.

Professor Whyte's research group at the Hunter Area Toxicology Service (HATS) in Newcastle, Australia, has maintained a prospective clinical database of all poisonings since 1987. The evidence from the HATS series of 2222 serotonergic overdoses has been published in a seminal series of papers [1–4], and there are many other recommended papers on the subject [5–12]. The typical clinical features of ST in humans, derived from the HATS data [4], are as follows:

  • • neuromuscular hyperactivity: tremor, clonus, myoclonus and hyperreflexia, and, in the advanced stage, pyramidal rigidity;
  • • autonomic hyperactivity: diaphoresis, fever, tachycardia, tachypnoea and mydriasis;
  • • altered mental status: agitation, excitement, with confusion in the advanced stage only.

Descriptions of the clinical presentation may be found elsewhere [1, 13]. Whyte concludes: ‘…only clonus (inducible, spontaneous or ocular), agitation, diaphoresis, tremor and hyperreflexia were needed for accurate prediction of ST as diagnosed by a clinical toxicologist.’ If, in the presence of a serotonergic agent, the single sign of spontaneous clonus is present, then ST may be reliably diagnosed.

The spectrum concept explains that ST is a progression of serotonergic effects mediated by the degree of elevation of intrasynaptic serotonin. These range from serotonin-related side-effects (at therapeutic doses) through to toxicity, and culminate in death with monoamine oxidase inhibitor/serotonin uptake inhibitor (MAOI/SRI) combinations. The three important mechanisms are serotonin reuptake inhibition, MAO inhibition and presynaptic release. The only therapeutic drugs implicated in severe reactions are overdose of MAOIs alone, combinations of MAOIs with either SRIs, or the only clinically available serotonin releaser, amphetamine (methylphenidate is not a risk [14]). The HATS data suggest that about 50% of patients who have ingested MAOIs (including the reversible inhibitor of monoamine oxidase type A (RIMA) moclobemide in combination with SRIs will exhibit at least moderately severe ST. The illicit drug MDMA (ecstasy; 3,4-methylenedioxymethamphetamine) is a serotonin releaser and may be lethal with any MAOI [15], but its effect is diminished by SRIs, because they impede its uptake into the presynaptic terminal, which prevents the releaser effect [16]. A case report contradicting that is worth writing!

SRIs alone do not result in severe ST or pyrexia in excess of 38.5 °C [2], which indicates that they have a ceiling effect. Understanding ST as a form of poisoning reveals the importance of knowing the degree to which different drugs are capable of elevating brain serotonin, because ‘the drug defines the syndrome’. Bayesian reasoning emphasises that ‘prior probability’ is the key [17] in assessing the likelihood and severity of ST [5], and, as Gill has pointed out [18], clinicians are natural Bayesians. Basic pharmacology reassures us that neuroleptic drugs cannot raise serotonin, ergo, they cannot produce ST, and NMS simply does not occur after SRIs because they do not deplete dopamine: yet both of these have been repeatedly and erroneously reported. Some of these reports have been counterbalanced in a series of comments emanating from Professor Whyte's group and the author [19–28]. However, once such reports are in the literature they are often quoted uncritically without the responses being noted. It is thus vital to have an accurate table of drugs with information about the degree to which they are capable of elevating serotonin, and, apart from tables published in journals, the author maintains such a list at http://www.psychotropical.com. The relative frequency and severity of ST with different drugs (and combinations) is useful in refining hypotheses about the actions of those drugs [29].

Death can result from a single dose of a drug (imipramine 225 mg) when errors are made in a patient taking an MAOI, as the recent Otte case in a European teaching hospital demonstrated [30]. Whilst it is fair to state that these authors did not make optimal use of their experience to increase their knowledge (as evidenced in their report, which did not entail systematic review), it seems unreasonable to suggest that all such doctors are unlearned and incompetent. Rather, it may be taken to indicate that clinical medicine is sufficiently complex to make it difficult for a great majority of doctors to absorb more information or make better real world decisions than they already do. Perhaps a continuing percentage of such outcomes are an inevitable consequence of the complexity of medicine and the human condition. Can case reports contribute to minimising this problem?

It is appropriate that all medical scientific work and publishing has been subjected to harsh criticisms recently as a result of the increasing appreciation of the pervasive bias and lack of objectivity that distorts the scientific method. It is probable that factors relating to commercial pressures to maintain the circulation of journals and to market pharmaceuticals are among the factors near the heart of these problems. A brief tour of comments made in eminent journals by senior members of the professions is necessary to create a perspective for readers. Smith, editor of the BMJ, ‘stepped down’ in 2004, and then wrote that ‘Medical journals are an extension of the marketing arm of pharmaceutical companies’[31]. He quoted Horton, the Lancet editor, as stating: ‘Journals have devolved into information laundering operations for the pharmaceutical industry’[32], and the NEJM editor: ‘primarily a marketing machine … co-opting every institution that might stand in its way’[33]. The titles of these pieces convey the tone. All doctors will benefit from some understanding of the existing ‘proof’ of unscientific and disingenuous practices engaged in manufacturing favourable evidence about drugs [34, 35], and if one imagines that it is reliable and objective to fall back on meta-analysis studies and practice guidelines, think again. The evidence is that they are frequently fatally tainted with bias, but of a more subtle and insidious kind [36, 37]. Case reports may be considered in the light of the recognition that systematically performed trials answer different questions, but may not be, in reality, more ‘objective’ or better real world science. Antidepressant drugs are still being ‘proved’ (a much used but misappropriated word) to work in trials, and then found not to work by ordinary clinicians [38].

Case reports represent a basic stage and form of scientific observation and can only address a limited and particular type of question; they generally serve the purpose of generating a hypothesis which then requires more systematically gathered observations to support or refute it [39]. However, the case report can, almost with a single instance, deliver the refutation represented by Popper's ‘one black swan’ (disproving the proposition ‘all swans are white’). They can also be just interesting and educational for students [40, 41]. Summaries of research methods [39] have examined case reports, case series and the reasons for avoiding them [42]. Infamous examples are given of how they can adversely influence clinical practice [42]. On the other hand, they are responsible for the recognition of many important therapeutic effects and adverse effects. Indeed, there is some irony in the observation that all the presently effective antidepressant drugs were discovered by astute clinical observation in case reports (often termed ‘serendipity’, although that is subtly dismissive), whereas there is no new antidepressant in 50 years that has been discovered because of systematically conducted research. So, in terms of results, case reports win 3–0! They are also responsible for the recognition of many important side-effects like the hypertensive reaction to cheese and numerous other side-effects, especially drug interactions.

Much of the above applies to ST which serves as a model for the vicissitudes of case reports throughout the 50-year history of the knowledge of this subject. Animal research defined the characteristics of ST around 1960 and systematic work in humans, not case reports, demonstrated the features of ST that the animal work had suggested [43]. Unfortunately, no-one learnt from history and many case reports throughout the 1970s and 1980s (that were unrecognised ST), were published [8]. Meanwhile, patients continued to die, largely because the pre-existing research was not incorporated into the corpus of medical knowledge and practice. It took 20 years before the connection with this pre-existing research was made in a report that is (still) widely and incorrectly quoted in most reviews as the first report of ST in humans [44].

This multitude of unfruitful, confusing and ill-informed reports continues. Few of them have been of any help, most have obfuscated the facts, confused the issues and perpetuated misinformation. My research of these case reports demonstrates that this has invariably been the result of failure to carry out adequate systematic reviews of the pre-existing literature, combined with an insufficient knowledge of the underlying body of science relating to the subject. This does not mean that case reports are inevitably useless, or that they should contain mini-reviews of the subject; indeed it would be better if they did not. It does indicate that editors and their referees are failing to ensure that a comprehensive knowledge and consideration of the pre-existing literature has been achieved and that relevant and representative references have been accurately and appropriately quoted. Once contributions are submitted to journals it is the responsibility of the editor and his appointed referees to maintain the standards of scientific literacy; they are the owners of the publication process and cannot abrogate this responsibility. There is much yet to be achieved, even in the ‘top’ journals. Surveys of journals' practices in relation to setting standards or guidelines for both authors and referees [45] demonstrate that more organisation, leadership and auditing of standards would be beneficial. The consequences are illustrated throughout the history of case reports on ST, which shows that there is much room for improvement. The BMJ's recent attempt to published ‘evidence based’ case reports [46] indicates that the difficulties are great, since despite this they have still produced one of the more egregious contributions of late [47]. The lion's share of the opportunities for change and progress lie with editors and referees [48] who are the owners of the publication business. An organisation for editors that can produce advice, assistance, co-ordination of policy and auditing of standards might catalyse change; meanwhile, individual editors may also take up the challenge.