The paper by Wong and colleagues1 in this issue of the journal, concluding that balloon angioplasty is preferred to surgery as the optimal treatment for patients with aortic coarctation, has been a source of extraordinarily extensive debate amongst the authors, the reviewers, the editor-in-chief, and his advisors. The manuscript has passed through several iterations, and discussions have centred not only on its contentious conclusion, but also on the potentially controversial methods through which the authors reached their conclusions. After much debate, it was agreed to publish the manuscript, so that you, the readers, could make up your own minds as to its validity, but to accompany the paper with the arguments that follow, presented in the form of an editorial commentary.
As you will quickly discover, the editorialists find two disparate themes are of central importance. We have, therefore, formulated the editorial around these two disparate themes. We anticipate that the authors may disagree with both issues, and we are more than ready to answer any rebuttal that they may wish to mount. The crux of the controversy, however, is what is this paper really about, and what does it tell us about medical decisions and real results?
The clinical decision-making process as driven by data
An ongoing central question is how to find scientific truth in clinical medicine, in full knowledge of its limitations that are functions mainly of era, culture, and affluence. As yet, no universal answer is possible to the question, but the following sources of information are available to us, in descending order of relevance:
• Systematic reviews and meta-analyses
• Randomised controlled trials
• Cohort studies
Wong and colleaguesReference Wong, Benson, Van Arsdell, Karamlou and McCrindle1 have made an extensive search of these categories of scientific evidence as published in the English language, and have found 104 papers. Of these, only two describe randomised trials, of 36Reference Cowley, Orsmond, Feola, McQuillan and Shaddy2 and 58Reference Hernandez-Gonzalez, Solorio and Conde-Carmona3 patients, respectively. It is clear that 2 papers are insufficient for a formal meta-analysis. For that reason, the authors have accumulated other literature, consisting of “case series”. Because of duplicate publication, 97 papers of the original 104 were used, of which 28 reported results of balloon angioplasty, and 68 of surgical repair.
We must ask, however, whether these papers give a reliable estimate of contemporary results of these modalities for treatment? Surgical repair of coarctation is one of the oldest operations in cardiac surgery. Crafoord performed the first coarctation repair on October 19, 1944!Reference Crafoord and Nylin4 The surgical results assessed by Wong and associatesReference Wong, Benson, Van Arsdell, Karamlou and McCrindle1 include series going back to the 1960’s! Clearly, results from that era, during which all kinds of partly outdated techniques were used, cannot seriously be compared to the results of contemporary balloon angioplasty. This mixed selection of published surgical results, nonetheless, was used as the basis for their investigation. So, although on first sight it seems reasonable to use all available literature, and the data it contains, in aggregated form as a basis for estimating results of the procedures under investigation, we are of the opinion that severe limitations are attached to indiscriminate use of cohort studies published over a period of 22 years.
We base our opinion, first, on the fact that these results are outdated. Second, although the period over which data was mined on surgical treatment of aortic coarctation was congruent with the period for which data was compiled about balloon angioplasty, follow-up is substantially different. The first balloon angioplasty for coarctation was reported in January, 1982,Reference Singer, Rowen and Dorsey5 and first reference dates from 1983, including no more than 4 patients,Reference Finley, Beaulieu, Nanton and Roy6 although Wong and colleaguesReference Wong, Benson, Van Arsdell, Karamlou and McCrindle1 claim to have used literature from 1984. The penetration of balloon angioplasty was limited during its early development. Thus, for all practical purposes, the periods over which patients were treated with surgery as opposed to angioplasty differ widely, with the surgical series starting some 20 years earlier. Median follow-up, therefore, is bound to differ grossly between the modalities, making comparison between them hazardous. Third, publication bias, alas, has proven “case series” or “cohort studies” provide unreliable evidence with which to estimate real results. More often than not, authors are inclined to report results that are favourable, and not bad results. The obvious answer is to use randomised trials, but as already discussed, only 2 such trials are available, one of which contains insufficient information with which to judge the reliability of the data.Reference Hernandez-Gonzalez, Solorio and Conde-Carmona3 This leaves one randomised trial, on which two papers have been published.Reference Cowley, Orsmond, Feola, McQuillan and Shaddy2, Reference Shaddy, Boucek and Sturtevant7 The bottom line of the latter paper is: “balloon angioplasty for the treatment of childhood aortic coarctation is associated with a higher incidence of aneurysm formation and iliofemoral artery injury than surgery. These differences should be considered when undertaking treatment for native aortic coarctation during childhood.”
The real results of therapeutic procedures such as those under debate might better be estimated through registries in which the participating medical professionals are anonymous. The threshold to report less favourable results is bound to be lower if there is no negative impact on the involved professionals. For that reason, we investigated the congenital cardiac database of the European Association for Cardiothoracic Surgery, which contains data on more than 51,000 procedures, of which 2,217 were repair of aortic coarctation. Data from 5 of the participating centres were audited each year, so it is possible to give an accurate estimate of contemporaneous results. Thus, audited results are available for 201 surgical repairs of aortic coarctation, with one death being recorded, yielding a morality of 0.5%. Of the total number of 2,217 repairs for aortic coarctation available in the audited and non-audited data sets, 10 deaths were recorded, yielding a mortality of 0.68%. The breakdown of these figures is detailed in the Table 1. As is shown in our Table, these results are substantially different from those presented in Table 2 of the paper from Wong and associates,Reference Wong, Benson, Van Arsdell, Karamlou and McCrindle1 where the mortality for surgical treatment is suggested to be 10% in neonates, 4% in children, and 3% in adolescents. Even if these results may have been influenced by the associated lesions, the distribution of which may well vary between data sets, current surgical mortality does not even begin to approach the figures cited in the paper of Wong and colleagues.1
The table shows the alleged mortality at different ages for surgical repair of aortic coarctation as cited by Wong and associates1 (1st column) compared to the results obtained from the database of the European Association for Cardiothoracic Surgery (ECDB). The figures show the number of deaths (+) compared to the number of procedures (N) for the total records (2nd and 3rd columns), and then to the audited procedures (4th and 5th columns).
We conclude that surgical results in the current era are excellent, with true mortality likely to be between 0.5% and 0.7% when considering all cases of isolated coarctation. Of those dying, the majority will almost certainly be made up of neonates.
The process of clinical decision-making as driven by opinion
A different perspective is that the quantitative aspects of the paper relating to the decision tree are of secondary importance. Rather, the central theme of the paper is that healthcare professionals making recommendations for treatment of aortic coarctation are driven by opinion, and not by data. Even a perfect surgical result would not sway the opinion of these healthcare workers who “prefer” balloon angioplasty, even if results are demonstrably worse. Readers should take careful note of the fact that the study does not assess the preferences of patients, or their parents, on whom the decision ultimately rests, and who are directly involved with the risks of either procedure. Thus, the paper is not about the “preferable” technique on the basis of outcomes, but about the preferences of healthcare workers who play a crucial role in parental decisions.
When viewed from this perspective, we submit that the title of the work is misleading. It gives the casual reader the impression that the paper will offer scientific evidence, driven by data, that balloon angioplasty “is to be preferred” to surgery, presumably because it has better outcomes. If that were the case, however, it makes no sense that the standard gamble was administered to healthcare workers, and not to parents of patients with aortic coarctation. We would argue that a more appropriate title could be “medical professionals prefer balloon angioplasty to surgery for aortic coarctation”. In fact, it might even be “medical professionals prefer balloon angioplasty to surgery for aortic coarctation, even if surgery can be performed with zero risk.” Or simply, “medical professionals are risk adverse when it comes to deciding between an invasive and less invasive procedure: don’t bother them with the facts.” The methodology of the standard gamble indicates that this preference would be unchanged, even if results of balloon angioplasty were proven to be inferior to surgery.Reference Kirklin and Blackstone8
Do we care what these medical professionals think? Note should also be taken of the fact that the medical professionals were 3 staff cardiologists, 5 fellows in paediatric cardiology, 4 cardiovascular research assistants, 3 experienced nurses from the paediatric cardiologic clinic, but zero cardiac surgeons, and zero interventional paediatric cardiologists! Why this mix? Are research assistants, nurses, or even fellows in paediatric cardiology instrumental in the decision-making process, as implied in the title of the manuscript? It may make sense to exclude surgeons, because the patient will never get to a surgeon if these healthcare workers can help it! Does it make sense to exclude interventionalists? It may if they, too, are excluded from the decision-making process.
What, therefore, do we learn from this paper? We learn that medical professionals, who are the gatekeepers of therapy, are so convinced that invasion of the body is to be avoided, almost at all costs, except perhaps in neonates, that they will rarely, if ever, refer a patient to surgery. This is likely of great interest! It would be of equally great interest to have had a parallel study of the true decision-makers, generally the parents of these babies. What level of evidence would it take for them to decide that invasion of the body is worthwhile, given a factual presentation of contemporary risks and benefits of surgical as opposed to percutaneous approaches to the repair of aortic coarctation? To find out would require a prospective study using the standard gamble approach. Only then would we know if the level of risk-aversion of caregivers equals that of parents faced with this difficult decision.
Given this perspective of the paper, we contend that there are a number of its contained features that could mislead the reader. If one reads the objective of the study, it does not state that it is the preferences of the medical professionals that are being studied. The casual reader may think the paper is addressing outcomes, just as the commentary above. As the reader goes on to the background of the study, it appears to indicate that it is difficult to make direct comparisons of alternative therapies for aortic coarctation with respect to outcome. But, no! What is difficult to assess is the attitudes of medical professionals regarding the two options for treatment. As readers seek to reach their own conclusions, they will encounter the word “preferred”, possibly without understanding that the authors have not addressed the preferred treatment of aortic coarctation as driven by contemporary data, but rather the preferences of a group of caregivers, having chosen to ignore the current data.
Why have the authors taken this approach? We found it impossible to clarify this point during the process of review. The authors state they have interviewed healthcare workers familiar with managing patients with aortic coarctation, and knowledgeable of outcomes for the two different approaches. These healthcare workers, however, are not the decision-makers. Medical professionals only make therapeutic recommendations. These recommendations, we would think, should be based on facts presented free of risk-aversion psychology, which might be translated into coercion, of the persons making the recommendation. Indeed, the claim that this psychology comes into play at all is the most important revelation of the paper. We find it of note, however, that this risk aversion is less for neonates than it is for infants and children.
If we return to the decision-makers, namely the parents of these individuals with aortic coarctation, it is of further interest that the authors of the paperReference Wong, Benson, Van Arsdell, Karamlou and McCrindle1 appear to have no faith in the ability of parents to understand the disease, its treatment, or results of the treatment, nor how to make an informed decision, despite the necessity of obtaining informed parental consent for intervention of either kind. They state, “It was not currently feasible to educate a large number of parents with regards to the pathophysiology, complications, and outcomes for aortic coarctation.” If the authors truly believe this, then they must also believe it is impossible to obtain informed parental consent. If so, we must ask if intervention of either kind for the treatment of aortic coarctation is currently being undertaken at their institution within the necessary framework of medical ethics?
