The strength of uncontrolled case series

In the hierarchy of clinical evidence uncontrolled case series occupy a lowly position, well below that of randomised trials. Is it a paradox therefore that most of the evidence of effective surgical treatments comes not from randomised trials but from uncontrolled case series? No, it is quite understandable, for randomised trials of surgical operations are notoriously difficult. Thus the design of a randomised trial can be impeccable – the statement of a hypothesis, a power calculation based on the hypothesis, the generation of the random allocation sequence, concealment of the treatment allocation to the point of surgery, and inclusion of all participants in the analysis of the trial – all that; but unless account is taken of factors which are difficult to quantify, such as the training of the operators and the difficulty of the operation, the trial is confounded and the results meaningless. Jeremy Hawe and his colleagues (pages 1083–1088) report an uncontrolled case series of three hundred women who had undergone laparoscopic-assisted Doderlein hysterectomy performed by eight fully trained surgeons. One-quarter of the women had endometriosis and one-fifth uterine fibroids. The average duration of the operation was 102 minutes and the average stay in hospital 80 hours. Four women sustained life-threatening complications, all of which were treated successfully. The strength of this case series is that it describes the outcomes of laparoscopic-assisted Doderlein hysterectomy as they typically occur when the operation is carried out by fully trained surgeons. The information in this uncontrolled case series is important to women and their gynaecologists in making decisions about hysterectomy. The authors were correct to resist the temptation to compare their results with historical or parallel controls, for it would be impossible to adjust for the biases inherent in such a comparison. The main uncertainty concerning laparoscopic-assisted Doderlein hysterectomy is its value compared with vaginal hysterectomy, but even here a randomised trial is unlikely to give information which is clinically useful. There would be difficulty in deciding the primary outcome; the size of trial based upon this outcome is likely to be very large; the trial may therefore involve many centres where the training of the operators is variable; and only women with minimal abnormalities in the pelvis would be included, such that the results of the trial could not be made general. This trial would therefore lack credibility and would not influence clinical practice. We should not despise uncontrolled case series.