Case reports of novel treatments - proper evaluation before clinical use

Abstract

The concept of evidence based medicine is based on the primacy of the randomised controlled trial (RCT) as the gold standard in evaluating new interventions. The hierarchy of evidence places a systematic review of RCTs with meta-analysis as the highest form of evidence and puts expert opinion at the bottom. This dogma has tended to devalue a variety of other methods of evaluating medical treatment, perhaps even denigrating the usefulness of expert opinion, consideration of observations and case reports. Case reports and case series continue to have an important role in the progress of our understanding of disease and its treatment, and can be vital in detecting novel or unexpected diseases or responses to treatment [1, 2]. They can provide new ideas and broaden our understanding of mechanisms as well as permit the discovery of unexpected effects, be they beneficial or adverse. However, they offer little insight into the magnitude or frequency of problems, both of which will have an effect on morbidity and mortality. There are many historically significant case reports in terms of disease, such as Silas Weir Mitchell’s reports on nerve injuries sustained in the American Civil War [3], Sigmund Freud’s numerous cases [4], Treves’s report of the Elephant Man [5] and, in more recent times, the initial case reports and series of patients who were later discovered to have been suffering from Acquired Immunodeficiency Syndrome [6]. New treatments have often been first described as case reports or series – for example, the first use of cocaine as a local anaesthetic [7]. If a new treatment appears to offer truly exceptional promise then its application can be taken up very rapidly. The initial demonstration of the effects of cocaine in the human eye was made on 11 September 1884 and presented at a meeting 4 days later [7]. By the end of November of that year, a case series of more than 150 patients operated on under topical anaesthesia of the eye was published [8]. This achievement, in an era of much slower communication, is perhaps one indication of the importance of the advance that had been discovered. Case reports also have value in drawing attention to complications that have previously been unrecognised, such as the early descriptions of malignant hyperpyrexia [9]. However, considerable caution must be exercised before the apparent benefits of a new treatment are accepted, and there are examples in the recent past when an early change in practice has proved unwise. Intravenous regional anaesthesia (Bier’s Block) was initially described in 1908 and, with the use of lidocaine or prilocaine, had an excellent safety record. One of the limitations of the technique was that a tourniquet was essential and operating time was limited by the duration that the patient could tolerate the tourniquet. In the 1980s there were suggestions from case reports that if bupivacaine was used as the local anaesthetic agent, particularly in high dosage, analgesia persisted after the tourniquet’s release and would allow surgery to continue. Perhaps the ready availability of bupivacaine and familiarity with its use contributed to an increase in the frequency with which this agent was employed. Although an early report suggested that this was safe with low concentrations of bupivacaine [10], the use of more concentrated solutions increased over time and was recommended by others [11]. The result is well known to most anaesthetists who were working in the UK in the 1980s. There were a number of deaths in accident and emergency departments where cardiac arrest ensued shortly after release of the cuff, particularly if this was early and accidental. Although initial comments were directed at the training and speciality of those administering the block, and the possibility of failure of double-cuff tourniquets, [12, 13] it was eventually realised that the underlying problem was not a change in either of these parameters but that an inappropriate drug was being used [14]. There may not even have been sufficient evidence that bupivacaine did indeed offer any significant benefit [15]. What change in clinical practice should occur, then, following the publication in this issue of Anaesthesia of a small case series purporting to show a very significant benefit of stellate ganglion block for postoperative analgesia [16]? Postoperative pain has assumed a high profile in the last two decades and any technique that offers significant advantages is likely to be seized upon rapidly. In considering the validity of observations, one should initially ask whether there is any credible basis for such a technique to provide benefit, and the scientific rationale and putative mechanisms are discussed in an accompanying editorial [17]. If that is the case, then the next step really must be a larger case series or observational study, then a formal RCT if the findings of such studies provide further justification. It is too easy to presume that because the apparent benefit is so great, there must be a physical effect, even if the mechanism is not fully understood. Postoperative pain is notoriously variable, and a useful study to bear in mind comes from the 1950s in the USA, where patients with intractable angina underwent bilateral internal mammary artery ligation in an attempt to control their symptoms. The results are quite impressive given the almost complete lack of any physiological basis for such a procedure [18]. It required more robust study designs to demonstrate the ineffectiveness of the surgery, the importance of the attitude of the attending physician and the placebo effect [18, 19]. McDonnell et al.’s study has been conducted in a highly reputable institution by experienced and respected clinicians, and there is nothing to question the integrity of their report. The authors’ final sentence, in which they caution that: ‘…these initial findings deserve further investigation’ is absolutely clear and correct. Stellate ganglion block, even in expert hands, can cause serious adverse effects and there is no good reason to introduce it immediately into routine clinical practice. It would be even less wise for a clinician to use this for the occasional case where postoperative pain is likely to be particularly troublesome. It would be all too easy for the method used by these authors to be adapted, e.g. by using larger doses or different drugs, by others who fail to reproduce their impressive findings. At the very least, this would expose patients to significant additional risk for as yet unproven benefit. For the present, it would be sensible to limit the use of this technique to well designed research studies, in selected groups of patients, in whom formal evaluation can be undertaken in controlled conditions and the results subjected to peer review with a view to publication in the public domain. No external funding and no competing interests declared.