Jump to comment:

• Provincial Letter - a reply to Shaw and Conway
  Raj Mohindra
  Published on: 18 May 2017
• Provincial Letter - a response to Shaw and Conway
  R Mohindra
  Published on: 18 May 2017
• Evidence, Equipoise & Endocarditis: Reply to Mohindra
  David M Shaw
  Published on: 18 May 2017

• Published on: 18 May 2017

  Provincial Letter - a reply to Shaw and Conway

  • Raj Mohindra

  1.Introduction

  The reply of Shaw and Conway[1] is certainly impassioned albeit somewhat diffuse. The diffuse nature is exposed by the fact that they make several arguments ad hominem e.g. the proposition that "Mohindra believes that logic should play no part in assessments of evidence" which is both an argument ad hominem and not predicated on any statements made. Although such arguments might have...

  Show More

  1.Introduction

  The reply of Shaw and Conway[1] is certainly impassioned albeit somewhat diffuse. The diffuse nature is exposed by the fact that they make several arguments ad hominem e.g. the proposition that "Mohindra believes that logic should play no part in assessments of evidence" which is both an argument ad hominem and not predicated on any statements made. Although such arguments might have rhetorical force they do not impact upon the substantive argument as a matter of logic. I will attend to the substantive arguments here. I do not regard the arguments ad hominem as sufficiently meritorious or relevant to justify more than passing attention.

  My paper in this journal is a focused discussion developing a substantive argument against the National Institute of Health and Clinical Excellence (NICE) guidance for the use of antibiotic prophylaxis to prevent infective endocarditis ('IE').[2] It is not a response to Shaw and Conway.

  Given the nature of the reply by Shaw and Conway I will address the substantive arguments that Shaw and Conway have offered their reply to my paper in this journal,[3] in their paper in the journal Heart[4] and their response to my letter in reply to their paper in Heart.[5] This will lengthen the discussion because the argument by Shaw and Conway is entirely distinct from my argument save that they both relate to the same NICE guidance. I apologise to the readership for this and for the delay in my response. I have been occupied with other projects.

  2.A key proposition in issue is empirical

  Although Shaw and Conway do not explicitly state their core proposition, it appears to be:

  SC proposition: antibiotic prophylaxis should not be used in patients undergoing invasive procedures (both dental and non-dental)

  In my paper a key proposition in issue is:

  RM proposition: Does antibiotic prophylaxis reduce the probability of infective endocarditis?

  Note that SC proposition is a normative assertion whilst RM proposition is an empirical proposition.

  The first fundamental problem with Shaw and Conway's argument is that they have failed to recognise that a critical proposition for their argument is an empirical one i.e. whether or not the use of antibiotic prophylaxis results actually works to reduce the risk of IE.

  The net benefit of antibiotic prophylaxis can be determined by evaluation of the net benefit inequality (see Equation 1)

  (harm caused by IE with antibiotic prophylaxis) + (harm caused by antibiotics) > [harm caused by IE without antibiotic prophylaxis]

  Equation 1: the net benefit equality (taken from my letter in response to Shaw and Conway's paper in Heart) The second term, the harm caused by antibiotics, includes all harm including risk of anaphylaxis[6] and population risk from any rise in bacterial resistance. It must do so or we would not be considering overall net benefit.

  If the net benefit inequality is true then, overall, more harm is caused with antibiotic prophylaxis than without it and we should not give antibiotic prophylaxis. If the net benefit inequality is false then more harm is caused without antibiotic prophylaxis and we should give antibiotic prophylaxis. If we cannot evaluate the net benefit inequality then we must look to reasons other than empirical ones upon which to choose whether or not to give antibiotic prophylaxis. Note that the value of the net benefit inequality is necessarily partly contingent on the value of RM proposition.

  The normative proposition SC proposition must rest on the fact that the net benefit inequality is not false or it cannot rest on an empirical basis. However, as we shall see below (see section 3), the net benefit inequality is presently essentially indeterminate. Given this SC proposition must rest on non- empirical premises. Any part of Shaw and Conway's argument that claims to rest on an empirical base must fail as a result.

  However Shaw and Conway's argument has gone beyond this and simply overlooked the fact that whether we should use antibiotic prophylaxis or not actually depends in great part upon whether or not antibiotic prophylaxis actually works to reduce the incidence of IE. This is evident in their view that the current analysis is best progressed by connecting the use or not of antibiotic prophylaxis with the outcome of IE or no IE. They make this connection in their paper in Heart when they deploy their cardiologist's wager (see Table 1) and their patient's wager, which is similar.

  	No infective endocarditis	Infective endocarditis develops
  Provide prophylaxis	Benefit	No loss
  Do not provide prophylaxis	No loss	Harm
  Table 1: the cardiologist's wager - taken from Shaw and Conway's paper in Heart[1]

  The problem is that the fact that antibiotic prophylaxis might reduce the risk of IE does not mean that everyone who gets antibiotic prophylaxis will necessarily not get IE.

  To see how this misconception infects Shaw and Conway's argument we can use an example. It is perfectly possible that in truth antibiotic prophylaxis carries negative net benefit overall (e.g. by significantly increasing the risk of developing IE through an increase in antibiotic resistance). If antibiotic prophylaxis was harmful overall but was still given we would regard this as the wrong thing to do even if the patient did not develop IE after receiving antibiotic prophylaxis. However looking at Shaw and Conway's cardiologist's wager in this situation, if no IE resulted after antibiotic prophylaxis was given the outcome would be regarded as a benefit.

  If Shaw and Conway's argument did take into account the impact of the net benefit inequality we would find that the cardiologist's wager and patient's wager in Shaw and Conway's paper would be replaced by the cardiologist's dilemma (Table 2) and patient's dilemma (Table 3) respectively. Note how the values in some of the boxes of these 2 matrices are indeterminable by virtue of lack of adequate empirical evidence.

  	Use antibiotic prophylaxis	Do not use antibiotic prophylaxis
  Antibiotic prophylaxis for IE carries net benefit overall	Save lives less the overall adverse effects of antibiotics	Lose lives
  Antibiotic prophylaxis for IE carries no net benefit overall	Side effects of antibiotics alone	No effect upon outcomes
  Table 2: The cardiologist's dilemma

  	Receive antibiotic prophylaxis	Do not receive antibiotic prophylaxis
  Antibiotic prophylaxis for IE carries net benefit overall	Good	Bad
  Antibiotic prophylaxis for IE carries no net benefit overall	Bad	Good
  Table 3: The patient's dilemma

  Alternatively, if Shaw and Conway argue that they are using the cardiologist's wager to illustrate the position they perceive is held by cardiologists then they are either guilty of setting up a straw man argument or they have a fundamental misconception of how cardiologist's think and thereby the other side of the argument. This latter point might be the explanation for the following further 2 fundamental misconceptions:

  1. that cardiologists base their decisions upon what is "best for the individual" "ignoring the wider population picture". As a cardiologist I would argue that I, and most of my colleagues, rest their argument upon their own view of the net benefit equality in the same way that Shaw and Conway appear to do. The problem is that, given the empirical uncertainty, more than one view of the value of the net benefit inequality is logically defensible and unfortunately, some of these logically defensible views are diametrically opposed; and

  2. that cardiologist's offer antibiotic prophylaxis is based upon "playing it safe" or a "no lose" approach. This is not correct because it is the view of the net benefit inequality rather than the operation of a precautionary principle[7] that ultimately determines whether or not one is in favour of giving antibiotic prophylaxis or not.

  Alternatively, if the "playing it safe" argument does bite against the cardiologist then it bites equally against Shaw and Conway because, as we will see below, Shaw and Conway's position is simply a mirror of the cardiologist's position differing only in their evaluation of the net benefit inequality.

  3.The evidence supporting the proposition that antibiotic prophylaxis reduces the incidence of infective endocarditis

  In relation to the standard of proof. RM proposition is about a clinical fact. As stated in my paper the generally accepted level of evidence required to reject the null hypothesis for propositions of clinical fact is sufficient clinical evidence to achieve a p value of <_0.05.8 unless="unless" this="this" level="level" of="of" certainty="certainty" is="is" achieved="achieved" in="in" one="one" direction="direction" or="or" another="another" the="the" truth="truth" value="value" and="and" magnitude="magnitude" rm="rm" proposition="proposition" remains="remains" undetermined="undetermined" only="only" controversially="controversially" determined="determined" because="because" we="we" cannot="cannot" reject="reject" null="null" hypothesis="hypothesis" relevant="relevant" direction.="direction." _="_" p="p"/>

  In their author's reply to my letter in Heart Shaw and Conway state "as we said in our paper, the evidence (though not conclusive) suggests that prophylaxis doesn't work". If Shaw and Conway wish to hold this position then the assertion that "[antibiotic] prophylaxis doesn't work" must be supported by sufficient empirical evidence to be able to reject the null hypothesis. If Shaw and Conway can establish this proposition to the standard of proof required to reject the null hypothesis on the basis of empirical data then they should do so.

  If they do not possess such data then they should accept, as they do within their own paper, NICE's conclusion that:

  "There is insufficient evidence to determine whether or not antibiotic prophylaxis in those at risk of developing infective endocarditis reduces the incidence of IE when given before a defined interventional procedure (both dental and non-dental)"

  In this context we should note that

  (1) in their own paper Shaw and Conway declare that "NICE's claim is not that prophylaxis does not work, but simply that there is insufficient evidence either way"; and

  (2) in their reply to my paper in this journal Shaw and Conway concede that "Mohindra is certainly right to point out that there is not much evidence regarding the benefits and harms of [antibiotic prophylaxis]."

  Further, in order to use Pascal's wager as an analogy to progress their own argument Shaw and Conway are required to accept that an imponderable or unproved fact lies at the heart of this debate.

  If there is insufficient evidence to conclude either RM proposition is true or that RM proposition is false then we have a state of equipoise. I would argue that the state of equipoise here is best characterised as insufficient evidence equipoise because the evidence base in support of each possible truth value of RM proposition does not reach the standard of proof required.

  4.The distinction between insufficient evidence equipoise and competing evidence equipoise

  As I point out in my paper in most situations there is no material difference between insufficient evidence equipoise and competing evidence equipoise. However here it is important because of the need to make a cost-effectiveness judgement. Evidence of the fact and degree of efficacy of the treatment is logically necessary before a cost-effectiveness judgement can be made.

  To restate this argument let us use an example. Start with an effectiveness judgement. In Table 4 let us say that, in relation to clinical effectiveness, NICE selects one truth value for each of treatment #1 and treatment #2. The choice it makes in each case is that it accepts that each respective treatment is efficacious (i.e. the left hand column).

  >

  	Evidence for efficacy	Evidence against efficacy
  Treatment #1	sufficient	sufficient
  Competing evidence equipoise
  Treatment #2	insufficient	insufficient	Insufficient evidence equipoise
  Table 4: Competing evidence equipoise vs Insufficient evidence equipoise

  Next move to the cost-effectiveness judgement. To do this NICE must look to the evidence that it has for efficacy of the treatment. For treatment #1 it has sufficient evidence to make a valid cost- effectiveness judgement. However for treatment #2 it cannot know how effective the treatment is or indeed whether the treatment is effective. Thus in cases of competing evidence equipoise NICE can make a valid cost-effectiveness judgement whilst in cases of insufficient evidence equipoise they cannot.

  This is one material difference between competing evidence equipoise and insufficient evidence equipoise. A key difference that lies not with the initial selection between truth values for deciding whether the treatment has efficacy or not. This selection is made in the face of equipoise (i.e. between 2 boxes of the matrix). The difference lies with the subsequent use of the available data to reach a cost-effectiveness judgement. This is because this second judgement rests solely upon one dataset (i.e. the contents of one box within the matrix).

  5.Why not omit to act if there is uncertainty?

  Shaw and Conway must explain why the omission to give antibiotics is different from the act of giving antibiotics in the context of a pre-existing legal duty of care.[9]

  Returning to Shaw and Conway's insistence that we should "simply avoid administering prophylaxis", the response must be "Why, given the position of insufficient evidence equipoise?"

  Shaw and Conway's try to answer this question using the following arguments inter alia:

  1. In their reply to my letter in Heart Shaw and Conway declare that "if there is really no evidence that an intervention works, we should not use it."

  This is Shaw and Conway's argumentum ad ignorantiam or appeal to ignorance.

  As I point out in my paper, the absence of evidence of benefit is not evidence of absence of benefit. Therefore ignorance of the effects of a treatment is not a conclusive reason not to use a treatment. This is particularly so where there is equipoise and the absence of evidence flows from a general paucity of evidence rather than a large body of available evidence demonstrating no net benefit.

  If there is a general paucity of evidence, as is the case here, then provided that there is insufficient evidence to conclude that the treatment in question does not carry net harm overall the normative conclusion "we should not use [antibiotic prophylaxis]" is contentious. This is because the argumentum ad ignorantiam lies available to both sides of the argument and carries equal force when deployed.

  As I point out in my paper, the better conclusion in a situation where there is insufficient evidence equipoise is that further research is necessary as a means by which to displace the null hypothesis.

  2. In their paper that Shaw and Conway declare that "NICE never claimed to have produced new evidence. They simply put an end to the practice of using an intervention for which there is no evidence base."

  The counter is that whilst there is no conclusive empirical evidence base upon which to continue the use of antibiotic prophylaxis, there is also no conclusive evidence base upon which to justify putting an end to the practice of using the antibiotic prophylaxis.

  3. Shaw and Conway argue that the overall net benefit must take into account the risk of anaphylaxis with antibiotics and the risk of increasing antibiotic resistance. In their paper Shaw and Conway ask us to accept that "if we have no evidence for the efficacy of prophylaxis (God's existence), should we not simply avoid administering prophylaxis (believing in Him) given the possible downsides in terms of anaphylaxis?"

  It is true, as Cardiologist's are aware all too keenly, that antibiotics can cause harm. Therefore in the net benefit inequality the 'harm caused by antibiotics' includes all the potential harms consequent upon the use of antibiotics. It is the net or 'all in' cost of antibiotic prophylaxis, not some subset of this, that is generally considered by cardiologists when evaluating the above inequality. However, given the presently available dataset, it is not possible to determine the overall net benefit inequality sufficiently well to be able to resolve the current debate.

  4. The fact that toothbrushing causes bacteraemia.

  Either antibiotic prophylaxis in patients undergoing invasive procedures (both dental and non-dental) reduces the incidence of IE or it does not. The fact of bacteraemia during toothbrushing does not impact upon the net benefit equality. For this reason it cannot move the argument here.[10] Whether or not we should give antibiotic prophylaxis for general toothbrushing is an entirely separate debate.

  5. Shaw and Conway declare that "it would be irresponsible to spend money on treatment that is not supported by evidence."

  As I argue in my paper, cost-effectiveness arguments cannot bite in the face of insufficient equipoise because they inevitably remain contentious. It is possible that the greater net cost lies with treating an increased number of patients with IE rather than with the use of antibiotic prophylaxis. Without knowing whether antibiotic prophylaxis carries net benefit or not we must remain uncertain as to cost- effectiveness because cost-effectiveness must be contingent upon knowledge of net benefit.

  None of these arguments effectively displaces the underlying insufficient evidence equipoise. Given this Shaw and Conway then move onto propositions that rest upon a purely non-empirical base in order to persuade. They aim to persuade predominately by attacking what they regard as the cardiologist's position.

  6.The non-empirical arguments

  Here Shaw and Conway declare that their argument is not theological but that they set out to demonstrate that "the attitudes towards evidence displayed by many cardiologists are similar to those of some religious people." They argue that "faith-based experimentation in the face of the evidence is exactly what our original paper targeted". In their paper they seek to do this by using an analogy between Pascal's wager and their cardiologist's wager.

  The structure of Pascal's wager is set out in Table 5:

  	Believe in God	Do not believe in God
  God exists	Win everything	Lose everything
  God does not exist	Lose nothing	Lose nothing
  Table 5: Structure of Pascal's wager

  Pascal's wager is an argument that seeks to give a reason to those who do not believe in the existence of God to believe in the existence of God. Pascal is clever because he does not rely on knowledge of the truth state of the underlying proposition i.e. does God exist? Pascal argues that believing in God means we could lose nothing or win everything whilst not believing in God means that we could lose everything or lose nothing. So, Pascal argues, if you must choose but cannot know whether God exists then why not choose to believe?

  Importantly this is not an argument that is about knowing. It is about choosing what to believe in the face of an imponderable fact. It is a theological argument in the sense that it is about choosing where to place one's faith.

  7.A theological argument?

  Given the overall lack of evidence a cardiologist may choose to believe that there is a net benefit in the use of antibiotic prophylaxis (let us assume that this is essentially a faith based argument given the evidence base). Based upon this view they would take the left hand column of the cardiologist's dilemma and use antibiotic prophylaxis.

  If Shaw and Conway follow NICE and take the right hand column of the same matrix they are faced with the same empirical dataset as the cardiologist and must make their choice based upon similarly non- empirical grounds. They may even find themselves believing that there is no net benefit in the use of antibiotic prophylaxis. But this would be as much a faith-based view as the cardiologist's position. If this is the basis of Shaw and Conway's argument then it is as theological (faith based) and as analogous to Pascal's wager as the cardiologist's position. The cardiologist and Shaw and Conway would simply be sitting on opposite sides of the same wager.

  If Shaw and Conway seek to escape this conclusion by arguing that they use some other non-empirical premise than belief for reaching their view then they should state that non-empirical premise. Whatever premise they choose to use the cardiologist is similarly entitled to use the same premise albeit with the converse truth value. In this way either both fall within the analogy with Pascal's wager or both do not.

  8.A bridge too far

  Shaw and Conway have attacked a weak cardiology position based upon insufficient empirical evidence from a similarly weak position also based upon insufficient empirical evidence. It is this that ultimately forces Shaw and Conway to use non-empirical premises to try and convince us to act in a particular direction especially after their appeal to ignorance fails.

  It was the fact that Shaw and Conway launched an untempered attack from such a weak empirical evidential base that was the source of my surprise. The structure of the debate, in particular the fact that the difference between the two sides rests essentially rests upon their respective views of the truth value of an empirically uncertain net benefit inequality, creates the conditions which leads to Shaw and Conway's own arguments being available to be simply reflected back upon them.

  I note that, by contrast, NICE was both measured and controlled in the development of their position. A position it sought to build on predominantly empirical premises.

  9.Why Pascal was right after all

  Shaw and Conway also appear to have fundamentally misunderstood the argument in my paper. They declare that "[Mohindra] seems to favour continuing to give patients a treatment for which there is no evidence because of the faith-based expert opinion of some cardiologists". I regard this proposition as, at best, a clear misperception of my argument and, at worst, an argument ad hominem. This is because, in my paper, I have been careful to craft my objection to the NICE guidance so as not to rest it upon any particular view of the empirical evidence base. I do not assert any truth value for net benefit inequality or assert that antibiotics should or should not be given. In this my argument stands distinct from the cardiologist straw man argument that Shaw and Conway have constructed.

  The conclusion of my paper makes this clear:

  "The argument here is not about how to interpret the available evidence. The objection is that the fact of insufficient evidence equipoise places the right to reach a cost-effectiveness decision outside the jurisdictional reach of NICE."

  10.Patients and paternalism

  The NICE guidance in this regard is as follows:[11]

  "Healthcare professionals should offer people at risk of infective endocarditis clear and consistent information about prevention, including ... the benefits and risks of antibiotic prophylaxis, and an explanation of why antibiotic prophylaxis is no longer routinely recommended"

  In their paper Shaw and Conway rhetorically ask "should dentists and doctors perhaps give patients the choice, stating clearly the relative risks of IE and antibiotics?" Unfortunately they do not appear to formally conclude 'yes' or 'no' to this question in their paper. They should make explicit their position here.

  If they conclude 'yes' then they must forgo the NICE guidance in favour of a sufficient explanation of the underlying insufficient evidence equipoise and then offer the patient a choice of whether or not to receive antibiotic prophylaxis. If the conclude 'no' then they are open to the charge of paternalism.

  The argument implicit in my paper is that because of the fact of insufficient evidence equipoise, as the ultimate bearer of risk, the patient should be informed and then be permitted a say in their choice of whether or not to receive antibiotic prophylaxis despite the complexity of the situation.

  11.Trials

  Shaw and Conway argue that "If anyone is guilty of conducting an unethical trial, it is cardiologists who gave [antibiotic prophylaxis] in the past or continue to do so."

  As I point out in my paper the difference between the pre-NICE guidance practice and the post-NICE guidance practice is not that the practice was universal, nor that the practice was undertaken despite the lack of an underlying clinical scientific evidence base. Each practice is open to challenge on these 2 grounds.

  The thing that makes the implementation of the NICE guidance a clinical trial is that NICE has systematically perturbed the pre-existing clinical practice in a way that can potentially provide empirical data which can be used to shed light on the truth value of RM proposition. Without this systematic perturbation of pre-existing practice no new knowledge could have been gained about RM proposition. The pre-existing practice alone could not deliver such new knowledge therefore it could not amount to de facto research.

  What makes this perturbation egregious is that NICE have done this despite their own guidance upon social value judgements and that this perturbation falls outside the national statutory framework for regulating clinical trials.

  In relation to the legal point on compensation it appears that Shaw and Conway are arguing a causation point in the tort of negligence. Such a point may arise in a legal case brought against NICE if the action is brought as a breach of statutory duty claim. I agree that an action in negligence is vulnerable on the causation point but it does not appear to be unarguable[12],[13],[14] although it would depend upon the underlying empirical evidence base. I would refer them to the section on causation in my Handbook of Medical Law[15] for further information.

  Importantly, for personal injury cases, the limitation period does not begin until it is first realised that a person has suffered a significant injury that may be attributable to the negligence of a third party.[16] If the data collection I suggest is required in my paper is undertaken and demonstrates a statistically significant increase in the incidence of IE since the implementation of the NICE guidance it would make it more possible for such legal actions to crystallise. My suggestion that data collection is mandated is partly grounded on this legal consequence.

  12.Ancillary arguments

  The words I use in the paper do not bear the interpretation that I implicitly assert that the NICE guidance was based upon "philosophical and theological argument". Within my paper I have summarised the arguments used by NICE to support their conclusions and I note that none of these rest on philosophical or theological premises. It is Shaw and Conway that have turned to philosophical and theological argument in order to persuade clinicians that they should follow the NICE guidance.

  I have no objection to the use of philosophical argument in a clinical journal however a distinction must be made between the rhetorical use of philosophical argument and philosophical argument itself. I would argue that where there is rhetorical use of philosophical argument such argument should be rhetorically balanced. If there is pure philosophical argument, e.g. the exploration of an important concept such as equipoise, then constructive criticism of the validity of the argument may be useful.

  The following propositions that are offered by Shaw and Conway need clarification:

  1. "logic is philosophy"

  2. "Criticising someone for using philosophical argument is the same as criticising him for using logical argument."

  These also appear to be propositions offered by Shaw and Conway which are deployed ad hominem and are not predicated on any statements made. I would note here that logic is a method of reasoning that, if followed, ensures that the conclusion of an argument is true provided that the premises of the argument are true. It has foundations in both philosophy and mathematics but also a place in law, scientific method and computing inter alia. The discipline of philosophy extends beyond the subject of logic.

  13.Conclusion

  I have found myself restating a number of arguments that are already present in my paper and letter to Heart although I welcome the opportunity to develop them in more detail.

  It remains the case that the empirical evidence base in clinical medicine is under continuing development. This means that the particular details of the present argument may fall redundant once the underlying empirical dataset is sufficiently enhanced. However the principles discussed will remain operational possibly in other areas.

  It is a pity that Shaw and Conway have felt the need to resort to argumentum ad hominem. However, given the quality of readership of this journal I have no doubt that they will recognise this line of argument for the logical fallacy that it is. In light of this I will leave the reader to conclude whether or not the above arguments have met the substantive points raised by Shaw and Conway.

  [1] Shaw D, Conway DI. Pascal's Wager, infective endocarditis and the "no-lose" philosophy in medicine. Heart 2010;96:15-18.

  [2]National Institute For Health And Clinical Excellence. Antimicrobial prophylaxis against infective endocarditis.

  http://guidance.nice.org.uk/CG64 (accessed 07/11/2010)

  [3]Mohindra RK. A case of insufficient evidence equipoise: the NICE guidance on antibiotic prophylaxis for the prevention of infective endocarditis. J Med Ethics 2010;36:567-570 doi:10.1136/jme.2010.036848.

  [4]Shaw D, Conway DI. Equipoise and endocarditis - Authors' reply. Heart 2010;96:1333. doi:10.1136/hrt.2010.200766

  [5]Mohindra R. NICE should not intervene where there is equipoise by virtue of insufficient evidence. Heart 2010;96:1333. doi:10.1136/hrt.2010.200758

  [6]Friedlander AH. .Antibiotic prophylaxis for dentistry is not associated with fatal anaphylaxis. Clin Med Res2010;8(2):79.Comment on: Clin Med Res 2009;7(3):63-8.

  [7] Cranor CF. Toward understanding aspects of the precautionary principle. J Med Philos 2004;29 (3):259-79.

  [8] Rawlins M. De testimonio: on the evidence for decisions about the use of therapeutic interventions. Lancet 2008;372:2152e61.

  [9]Airedale NHS Trust v. Bland[1992] UKHL 5

  [10]Ignoratio elenchi- this is an interesting and apparently true fact but irrelevant to the determination of the empirical fact in issue.

  [11]National Institute For Health And Clinical Excellence. Antimicrobial prophylaxis against infective endocarditis. http://guidance.nice.org.uk/CG64 at Para 1.12:

  [12] McGheev National Coal Board [1972] 3 All ER 1008, HL

  [13]Bonnington Castings Ltd v Wardlaw [1956] 1 All ER 615, HL

  [14]Bolitho v City & Hackney HA [1997] UKHL 46, HL

  [15] Mohindra R. Handbook of Medical Law. Radcliffe Publishing (2008). ISBN-10 1846190673 ISBN-13 9781846190674 (296 pages)

  [16]S.11(4) Limitation Act 1980

  Conflict of Interest: None declared

  Show Less

  Conflict of Interest:
  None declared.

  • Back to top
• Published on: 18 May 2017

  Provincial Letter - a response to Shaw and Conway

  • R Mohindra, physician

  1.Introduction

  The reply of Shaw and Conway[1] is certainly impassioned albeit somewha...

  Show More

  1.Introduction

  The reply of Shaw and Conway[1] is certainly impassioned albeit somewhat diffuse. The diffuse nature is exposed by the fact that they make several arguments ad hominem e.g. the proposition that "Mohindra believes that logic should play no part in assessments of evidence" which is both an argument ad hominem and not predicated on any statements made. Although such arguments might have rhetorical force they do not impact upon the substantive argument as a matter of logic. I will attend to the substantive arguments here. I do not regard the arguments ad hominem as sufficiently meritorious or relevant to justify more than passing attention.

  My paper in this journal is a focused discussion developing a substantive argument against the National Institute of Health and Clinical Excellence (NICE) guidance for the use of antibiotic prophylaxis to prevent infective endocarditis ('IE').[2] It is not a response to Shaw and Conway.

  Given the nature of the reply by Shaw and Conway I will address the substantive arguments that Shaw and Conway have offered their reply to my paper in this journal,[3] in their paper in the journal Heart[4]and their response to my letter in reply to their paper in Heart.[5] This will lengthen the discussion because the argument by Shaw and Conway is entirely distinct from my argument save that they both relate to the same NICE guidance. I apologise to the readership for this and for the delay in my response. I have been occupied with other projects.

  2.A key proposition in issue is empirical

  Although Shaw and Conway do not explicitly state their core proposition, it appears to be:

  SC proposition: antibiotic prophylaxis should not be used in patients undergoing invasive procedures (both dental and non-dental)

  In my paper a key proposition in issue is:

  RM proposition: Does antibiotic prophylaxis reduce the probability of infective endocarditis?

  Note that SC proposition is a normative assertion whilst RM proposition is an empirical proposition.

  The first fundamental problem with Shaw and Conway's argument is that they have failed to recognise that a critical proposition for their argument is an empirical one i.e. whether or not the use of antibiotic prophylaxis results actually works to reduce the risk of IE.

  The net benefit of antibiotic prophylaxis can be determined by evaluation of the net benefit inequality (see Equation 1)

  (harm caused by IE with antibiotic prophylaxis)	+	(harm caused by antibiotics)	>	[harm caused by IE without antibiotic prophylaxis]
  Equation 1: the net benefit equality (taken from my letter in response to Shaw and Conway's paper in Heart) The second term, the harm caused by antibiotics, includes all harm including risk of anaphylaxis[6] and population risk from any rise in bacterial resistance. It must do so or we would not be considering overall net benefit.

  If the net benefit inequality is true then, overall, more harm is caused with antibiotic prophylaxis than without it and we should not give antibiotic prophylaxis. If the net benefit inequality is false then more harm is caused without antibiotic prophylaxis and we should give antibiotic prophylaxis. If we cannot evaluate the net benefit inequality then we must look to reasons other than empirical ones upon which to choose whether or not to give antibiotic prophylaxis. Note that the value of the net benefit inequality is necessarily partly contingent on the value of RM proposition.

  The normative proposition SC proposition must rest on the fact that the net benefit inequality is not false or it cannot rest on an empirical basis. However, as we shall see below (see section 3), the net benefit inequality is presently essentially indeterminate. Given this SC proposition must rest on non-empirical premises. Any part of Shaw and Conway's argument that claims to rest on an empirical base must fail as a result.

  However Shaw and Conway's argument has gone beyond this and simply overlooked the fact that whether we should use antibiotic prophylaxis or not actually depends in great part upon whether or not antibiotic prophylaxis actually works to reduce the incidence of IE. This is evident in their view that the current analysis is best progressed by connecting the use or not of antibiotic prophylaxis with the outcome of IE or no IE. They make this connection in their paper in Heart when they deploy their cardiologist's wager (see Table 1) and their patient's wager, which is similar.

  	No infective endocarditis	Infective endocarditis develops
  Provide prophylaxis	Benefit	No loss
  Do not provide prophylaxis	No loss	Harm
  Table 1: the cardiologist's wager - taken from Shaw and Conway's paper in Heart1

  The problem is that the fact that antibiotic prophylaxis might reduce the risk of IE does not mean that everyone who gets antibiotic prophylaxis will necessarily not get IE.

  To see how this misconception infects Shaw and Conway's argument we can use an example. It is perfectly possible that in truth antibiotic prophylaxis carries negative net benefit overall (e.g. by significantly increasing the risk of developing IE through an increase in antibiotic resistance). If antibiotic prophylaxis was harmful overall but was still given we would regard this as the wrong thing to do even if the patient did not develop IE after receiving antibiotic prophylaxis. However looking at Shaw and Conway's cardiologist's wager in this situation, if no IE resulted after antibiotic prophylaxis was given the outcome would be regarded as a benefit.

  If Shaw and Conway's argument did take into account the impact of the net benefit inequality we would find that the cardiologist's wager and patient's wager in Shaw and Conway's paper would be replaced by the cardiologist's dilemma (Table 2) and patient's dilemma (Table 3) respectively. Note how the values in some of the boxes of these 2 matrices are indeterminable by virtue of lack of adequate empirical evidence.

  	Use antibiotic prophylaxis	Do not use antibiotic prophylaxis
  Antibiotic prophylaxis for IE carries net benefit overall	Save lives less the overall adverse effects of antibiotics	Lose lives
  Antibiotic prophylaxis for IE carries no net benefit overall	Side effects of antibiotics alone	No effect upon outcomes
  Table 2: The cardiologist's dilemma

  	Receive antibiotic prophylaxis	Do not receive antibiotic prophylaxis
  Antibiotic prophylaxis for IE carries net benefit overall	good	bad
  Antibiotic prophylaxis for IE carries no net benefit overall	bad	good
  Table 3: The patient's dilemma

  Alternatively, if Shaw and Conway argue that they are using the cardiologist's wager to illustrate the position they perceive is held by cardiologists then they are either guilty of setting up a straw man argument or they have a fundamental misconception of how cardiologist's think and thereby the other side of the argument. This latter point might be the explanation for the following further 2 fundamental misconceptions:

  1. that cardiologists base their decisions upon what is "best for the individual" "ignoring the wider population picture". As a cardiologist I would argue that I, and most of my colleagues, rest their argument upon their own view of the net benefit equality in the same way that Shaw and Conway appear to do. The problem is that, given the empirical uncertainty, more than one view of the value of the net benefit inequality is logically defensible and unfortunately, some of these logically defensible views are diametrically opposed; and

  2. that cardiologist's offer antibiotic prophylaxis is based upon "playing it safe" or a "no lose" approach. This is not correct because it is the view of the net benefit inequality rather than the operation of a precautionary principle[7] that ultimately determines whether or not one is in favour of giving antibiotic prophylaxis or not.

  Alternatively, if the "playing it safe" argument does bite against the cardiologist then it bites equally against Shaw and Conway because, as we will see below, Shaw and Conway's position is simply a mirror of the cardiologist's position differing only in their evaluation of the net benefit inequality.

  3.The evidence supporting the proposition that antibiotic prophylaxis reduces the incidence of infective endocarditis

  In relation to the standard of proof. RM proposition is about a clinical fact. As stated in my paper the generally accepted level of evidence required to reject the null hypothesis for propositions of clinical fact is sufficient clinical evidence to achieve a p value of <0.05.[8] Unless this level of certainty is achieved in one direction or another the truth value and magnitude of RM proposition remains undetermined or only controversially determined because we cannot reject the null hypothesis in the relevant direction.

  In their author's reply to my letter in Heart Shaw and Conway state "as we said in our paper, the evidence (though not conclusive) suggests that prophylaxis doesn't work". If Shaw and Conway wish to hold this position then the assertion that "[antibiotic] prophylaxis doesn't work" must be supported by sufficient empirical evidence to be able to reject the null hypothesis. If Shaw and Conway can establish this proposition to the standard of proof required to reject the null hypothesis on the basis of empirical data then they should do so.

  If they do not possess such data then they should accept, as they do within their own paper, NICE's conclusion that:

  "There is insufficient evidence to determine whether or not antibiotic prophylaxis in those at risk of developing infective endocarditis reduces the incidence of IE when given before a defined interventional procedure (both dental and non-dental)"

  In this context we should note that

  (1) in their own paper Shaw and Conway declare that "NICE's claim is not that prophylaxis does not work, but simply that there is insufficient evidence either way"; and

  (2) in their reply to my paper in this journal Shaw and Conway concede that "Mohindra is certainly right to point out that there is not much evidence regarding the benefits and harms of [antibiotic prophylaxis]."

  Further, in order to use Pascal's wager as an analogy to progress their own argument Shaw and Conway are required to accept that an imponderable or unproved fact lies at the heart of this debate.

  If there is insufficient evidence to conclude either RM proposition is true or that RM proposition is false then we have a state of equipoise. I would argue that the state of equipoise here is best characterised as insufficient evidence equipoise because the evidence base in support of each possible truth value of RM proposition does not reach the standard of proof required.

  4.The distinction between insufficient evidence equipoise and competing evidence equipoise

  As I point out in my paper in most situations there is no material difference between insufficient evidence equipoise and competing evidence equipoise. However here it is important because of the need to make a cost-effectiveness judgement. Evidence of the fact and degree of efficacy of the treatment is logically necessary before a cost-effectiveness judgement can be made.

  To restate this argument let us use an example. Start with an effectiveness judgement. In Table 4 let us say that, in relation to clinical effectiveness, NICE selects one truth value for each of treatment #1 and treatment #2. The choice it makes in each case is that it accepts that each respective treatment is efficacious (i.e. the left hand column).

  	Evidence for efficacy	Evidence against efficacy
  Treatment #1	sufficient	sufficient	Competing evidence equipoise
  Treatment #2	insufficient	insufficient	Insufficient evidence equipoise
  Table 4: Competing evidence equipoise vs Insufficient evidence equipoise

  Next move to the cost-effectiveness judgement. To do this NICE must look to the evidence that it has for efficacy of the treatment. For treatment #1 it has sufficient evidence to make a valid cost-effectiveness judgement. However for treatment #2 it cannot know how effective the treatment is or indeed whether the treatment is effective. Thus in cases of competing evidence equipoise NICE can make a valid cost-effectiveness judgement whilst in cases of insufficient evidence equipoise they cannot.

  This is one material difference between competing evidence equipoise and insufficient evidence equipoise. A key difference that lies not with the initial selection between truth values for deciding whether the treatment has efficacy or not. This selection is made in the face of equipoise (i.e. between 2 boxes of the matrix). The difference lies with the subsequent use of the available data to reach a cost-effectiveness judgement. This is because this second judgement rests solely upon one dataset (i.e. the contents of one box within the matrix).

  5.Why not omit to act if there is uncertainty?

  Shaw and Conway must explain why the omission to give antibiotics is different from the act of giving antibiotics in the context of a pre-existing legal duty of care.[9]

  Returning to Shaw and Conway's insistence that we should "simply avoid administering prophylaxis", the response must be "Why, given the position of insufficient evidence equipoise?"

  Shaw and Conway's try to answer this question using the following arguments inter alia:

  1. In their reply to my letter in Heart Shaw and Conway declare that "if there is really no evidence that an intervention works, we should not use it."

  This is Shaw and Conway's argumentum ad ignorantiam or appeal to ignorance.

  As I point out in my paper, the absence of evidence of benefit is not evidence of absence of benefit. Therefore ignorance of the effects of a treatment is not a conclusive reason not to use a treatment. This is particularly so where there is equipoise and the absence of evidence flows from a general paucity of evidence rather than a large body of available evidence demonstrating no net benefit.

  If there is a general paucity of evidence, as is the case here, then provided that there is insufficient evidence to conclude that the treatment in question does not carry net harm overall the normative conclusion "we should not use [antibiotic prophylaxis]" is contentious. This is because the argumentum ad ignorantiam lies available to both sides of the argument and carries equal force when deployed.

  As I point out in my paper, the better conclusion in a situation where there is insufficient evidence equipoise is that further research is necessary as a means by which to displace the null hypothesis.

  2. In their paper that Shaw and Conway declare that "NICE never claimed to have produced new evidence. They simply put an end to the practice of using an intervention for which there is no evidence base."

  The counter is that whilst there is no conclusive empirical evidence base upon which to continue the use of antibiotic prophylaxis, there is also no conclusive evidence base upon which to justify putting an end to the practice of using the antibiotic prophylaxis.

  3. Shaw and Conway argue that the overall net benefit must take into account the risk of anaphylaxis with antibiotics and the risk of increasing antibiotic resistance. In their paper Shaw and Conway ask us to accept that "if we have no evidence for the efficacy of prophylaxis (God's existence), should we not simply avoid administering prophylaxis (believing in Him) given the possible downsides in terms of anaphylaxis?"

  It is true, as Cardiologist's are aware all too keenly, that antibiotics can cause harm. Therefore in the net benefit inequality the 'harm caused by antibiotics' includes all the potential harms consequent upon the use of antibiotics. It is the net or 'all in' cost of antibiotic prophylaxis, not some subset of this, that is generally considered by cardiologists when evaluating the above inequality. However, given the presently available dataset, it is not possible to determine the overall net benefit inequality sufficiently well to be able to resolve the current debate.

  4. The fact that toothbrushing causes bacteraemia.

  Either antibiotic prophylaxis in patients undergoing invasive procedures (both dental and non-dental) reduces the incidence of IE or it does not. The fact of bacteraemia during toothbrushing does not impact upon the net benefit equality. For this reason it cannot move the argument here.[10] Whether or not we should give antibiotic prophylaxis for general toothbrushing is an entirely separate debate.

  5. Shaw and Conway declare that "it would be irresponsible to spend money on treatment that is not supported by evidence."

  As I argue in my paper, cost-effectiveness arguments cannot bite in the face of insufficient equipoise because they inevitably remain contentious. It is possible that the greater net cost lies with treating an increased number of patients with IE rather than with the use of antibiotic prophylaxis. Without knowing whether antibiotic prophylaxis carries net benefit or not we must remain uncertain as to cost-effectiveness because cost-effectiveness must be contingent upon knowledge of net benefit.

  None of these arguments effectively displaces the underlying insufficient evidence equipoise. Given this Shaw and Conway then move onto propositions that rest upon a purely non-empirical base in order to persuade. They aim to persuade predominately by attacking what they regard as the cardiologist's position.

  6.The non-empirical arguments

  Here Shaw and Conway declare that their argument is not theological but that they set out to demonstrate that "the attitudes towards evidence displayed by many cardiologists are similar to those of some religious people." They argue that "faith-based experimentation in the face of the evidence is exactly what our original paper targeted". In their paper they seek to do this by using an analogy between Pascal's wager and their cardiologist's wager.

  The structure of Pascal's wager is set out in Table 5:

  	Believe in God	Do not believe in God
  God exists	Win everything	Lose everything
  God does not exist	Lose nothing	Lose nothing
  Table 5: Structure of Pascal's wager

  Pascal's wager is an argument that seeks to give a reason to those who do not believe in the existence of God to believe in the existence of God. Pascal is clever because he does not rely on knowledge of the truth state of the underlying proposition i.e. does God exist? Pascal argues that believing in God means we could lose nothing or win everything whilst not believing in God means that we could lose everything or lose nothing. So, Pascal argues, if you must choose but cannot know whether God exists then why not choose to believe?

  Importantly this is not an argument that is about knowing. It is about choosing what to believe in the face of an imponderable fact. It is a theological argument in the sense that it is about choosing where to place one's faith.

  7.A theological argument?

  Given the overall lack of evidence a cardiologist may choose to believe that there is a net benefit in the use of antibiotic prophylaxis (let us assume that this is essentially a faith based argument given the evidence base). Based upon this view they would take the left hand column of the cardiologist's dilemma and use antibiotic prophylaxis.

  If Shaw and Conway follow NICE and take the right hand column of the same matrix they are faced with the same empirical dataset as the cardiologist and must make their choice based upon similarly non-empirical grounds. They may even find themselves believing that there is no net benefit in the use of antibiotic prophylaxis. But this would be as much a faith-based view as the cardiologist's position. If this is the basis of Shaw and Conway's argument then it is as theological (faith based) and as analogous to Pascal's wager as the cardiologist's position. The cardiologist and Shaw and Conway would simply be sitting on opposite sides of the same wager.

  If Shaw and Conway seek to escape this conclusion by arguing that they use some other non-empirical premise than belief for reaching their view then they should state that non-empirical premise. Whatever premise they choose to use the cardiologist is similarly entitled to use the same premise albeit with the converse truth value. In this way either both fall within the analogy with Pascal's wager or both do not.

  8.A bridge too far

  Shaw and Conway have attacked a weak cardiology position based upon insufficient empirical evidence from a similarly weak position also based upon insufficient empirical evidence. It is this that ultimately forces Shaw and Conway to use non-empirical premises to try and convince us to act in a particular direction especially after their appeal to ignorance fails.

  It was the fact that Shaw and Conway launched an untempered attack from such a weak empirical evidential base that was the source of my surprise. The structure of the debate, in particular the fact that the difference between the two sides rests essentially rests upon their respective views of the truth value of an empirically uncertain net benefit inequality, creates the conditions which leads to Shaw and Conway's own arguments being available to be simply reflected back upon them.

  I note that, by contrast, NICE was both measured and controlled in the development of their position. A position it sought to build on predominantly empirical premises.

  9.Why Pascal was right after all

  Shaw and Conway also appear to have fundamentally misunderstood the argument in my paper. They declare that "[Mohindra] seems to favour continuing to give patients a treatment for which there is no evidence because of the faith-based expert opinion of some cardiologists". I regard this proposition as, at best, a clear misperception of my argument and, at worst, an argument ad hominem. This is because, in my paper, I have been careful to craft my objection to the NICE guidance so as not to rest it upon any particular view of the empirical evidence base. I do not assert any truth value for net benefit inequality or assert that antibiotics should or should not be given. In this my argument stands distinct from the cardiologist straw man argument that Shaw and Conway have constructed.

  The conclusion of my paper makes this clear:

  "The argument here is not about how to interpret the available evidence. The objection is that the fact of insufficient evidence equipoise places the right to reach a cost-effectiveness decision outside the jurisdictional reach of NICE."

  10.Patients and paternalism

  The NICE guidance in this regard is as follows:[11]

  "Healthcare professionals should offer people at risk of infective endocarditis clear and consistent information about prevention, including ... the benefits and risks of antibiotic prophylaxis, and an explanation of why antibiotic prophylaxis is no longer routinely recommended"

  In their paper Shaw and Conway rhetorically ask "should dentists and doctors perhaps give patients the choice, stating clearly the relative risks of IE and antibiotics?" Unfortunately they do not appear to formally conclude 'yes' or 'no' to this question in their paper. They should make explicit their position here.

  If they conclude 'yes' then they must forgo the NICE guidance in favour of a sufficient explanation of the underlying insufficient evidence equipoise and then offer the patient a choice of whether or not to receive antibiotic prophylaxis. If the conclude 'no' then they are open to the charge of paternalism.

  The argument implicit in my paper is that because of the fact of insufficient evidence equipoise, as the ultimate bearer of risk, the patient should be informed and then be permitted a say in their choice of whether or not to receive antibiotic prophylaxis despite the complexity of the situation.

  11.Trials

  Shaw and Conway argue that "If anyone is guilty of conducting an unethical trial, it is cardiologists who gave [antibiotic prophylaxis] in the past or continue to do so."

  As I point out in my paper the difference between the pre-NICE guidance practice and the post-NICE guidance practice is not that the practice was universal, nor that the practice was undertaken despite the lack of an underlying clinical scientific evidence base. Each practice is open to challenge on these 2 grounds.

  The thing that makes the implementation of the NICE guidance a clinical trial is that NICE has systematically perturbed the pre-existing clinical practice in a way that can potentially provide empirical data which can be used to shed light on the truth value of RM proposition. Without this systematic perturbation of pre-existing practice no new knowledge could have been gained about RM proposition. The pre-existing practice alone could not deliver such new knowledge therefore it could not amount to de facto research.

  What makes this perturbation egregious is that NICE have done this despite their own guidance upon social value judgements and that this perturbation falls outside the national statutory framework for regulating clinical trials.

  In relation to the legal point on compensation it appears that Shaw and Conway are arguing a causation point in the tort of negligence. Such a point may arise in a legal case brought against NICE if the action is brought as a breach of statutory duty claim. I agree that an action in negligence is vulnerable on the causation point but it does not appear to be unarguable[12]^,[13],[14] although it would depend upon the underlying empirical evidence base. I would refer them to the section on causation in my Handbook of Medical Law[15] for further information.

  Importantly, for personal injury cases, the limitation period does not begin until it is first realised that a person has suffered a significant injury that may be attributable to the negligence of a third party.[16] If the data collection I suggest is required in my paper is undertaken and demonstrates a statistically significant increase in the incidence of IE since the implementation of the NICE guidance it would make it more possible for such legal actions to crystallise. My suggestion that data collection is mandated is partly grounded on this legal consequence.

  12.Ancillary arguments

  The words I use in the paper do not bear the interpretation that I implicitly assert that the NICE guidance was based upon "philosophical and theological argument". Within my paper I have summarised the arguments used by NICE to support their conclusions and I note that none of these rest on philosophical or theological premises. It is Shaw and Conway that have turned to philosophical and theological argument in order to persuade clinicians that they should follow the NICE guidance.

  I have no objection to the use of philosophical argument in a clinical journal however a distinction must be made between the rhetorical use of philosophical argument and philosophical argument itself. I would argue that where there is rhetorical use of philosophical argument such argument should be rhetorically balanced. If there is pure philosophical argument, e.g. the exploration of an important concept such as equipoise, then constructive criticism of the validity of the argument may be useful.

  The following propositions that are offered by Shaw and Conway need clarification:

  1. "logic is philosophy"

  2. "Criticising someone for using philosophical argument is the same as criticising him for using logical argument."

  These also appear to be propositions offered by Shaw and Conway which are deployed ad hominem and are not predicated on any statements made. I would note here that logic is a method of reasoning that, if followed, ensures that the conclusion of an argument is true provided that the premises of the argument are true. It has foundations in both philosophy and mathematics but also a place in law, scientific method and computing inter alia. The discipline of philosophy extends beyond the subject of logic.

  13.Conclusion

  I have found myself restating a number of arguments that are already present in my paper and letter to Heart although I welcome the opportunity to develop them in more detail.

  It remains the case that the empirical evidence base in clinical medicine is under continuing development. This means that the particular details of the present argument may fall redundant once the underlying empirical dataset is sufficiently enhanced. However the principles discussed will remain operational possibly in other areas.

  It is a pity that Shaw and Conway have felt the need to resort to argumentum ad hominem. However, given the quality of readership of this journal I have no doubt that they will recognise this line of argument for the logical fallacy that it is. In light of this I will leave the reader to conclude whether or not the above arguments have met the substantive points raised by Shaw and Conway.

  
  

  ---

  [1] Shaw D, Conway DI. Pascal's Wager, infective endocarditis and the "no-lose" philosophy in medicine. Heart 2010;96:15-18.

  [2]National Institute For Health And Clinical Excellence. Antimicrobial prophylaxis against infective endocarditis.

  http://guidance.nice.org.uk/CG64 (accessed 07/11/2010)

  [3]Mohindra RK. A case of insufficient evidence equipoise: the NICE guidance on antibiotic prophylaxis for the prevention of infective endocarditis. J Med Ethics 2010;36:567-570 doi:10.1136/jme.2010.036848.

  [4]Shaw D, Conway DI. Equipoise and endocarditis - Authors' reply. Heart 2010;96:1333. doi:10.1136/hrt.2010.200766

  [5]Mohindra R. NICE should not intervene where there is equipoise by virtue of insufficient evidence. Heart 2010;96:1333. doi:10.1136/hrt.2010.200758

  ^[6]Friedlander AH. .Antibiotic prophylaxis for dentistry is not associated with fatal anaphylaxis. Clin Med Res2010;8(2):79.Comment on: Clin Med Res 2009;7(3):63-8.

  [7] Cranor CF. Toward understanding aspects of the precautionary principle. J Med Philos 2004;29(3):259-79.

  [8] Rawlins M. De testimonio: on the evidence for decisions about the use of therapeutic interventions. Lancet 2008;372:2152e61.

  [9]Airedale NHS Trust v. Bland[1992] UKHL 5

  [10]Ignoratio elenchi- this is an interesting and apparently true fact but irrelevant to the determination of the empirical fact in issue.

  [11]National Institute For Health And Clinical Excellence. Antimicrobial prophylaxis against infective endocarditis. http://guidance.nice.org.uk/CG64 at Para 1.12:

  [12] McGheev National Coal Board [1972] 3 All ER 1008, HL

  [13]Bonnington Castings Ltd v Wardlaw [1956] 1 All ER 615, HL

  [14]Bolitho v City & Hackney HA [1997] UKHL 46, HL

  [15] Mohindra R. Handbook of Medical Law. Radcliffe Publishing (2008). ISBN-10 1846190673 ISBN-13 9781846190674 (296 pages)

  [16]S.11(4) Limitation Act 1980

  Conflict of Interest:

  None declared

  Show Less

  Conflict of Interest:
  None declared.

  • Back to top
• Published on: 18 May 2017

  Evidence, Equipoise & Endocarditis: Reply to Mohindra

  • David M Shaw, Lecturer in Ethics
  • Other Contributors:
    • David I Conway

  Introduction

  RK Mohindra's recent paper criticises NICE's decision-making regarding the use of antibiotic prophylaxis to prevent infective endocarditis. He is also critical of our defence of NICE, but does not actually engage with our argument, stating simply that he was "surprised" to see a philosophical defence of something that should be decided by evidence. We find it surprising that Mohindra believes that...

  Show More

  Introduction

  RK Mohindra's recent paper criticises NICE's decision-making regarding the use of antibiotic prophylaxis to prevent infective endocarditis. He is also critical of our defence of NICE, but does not actually engage with our argument, stating simply that he was "surprised" to see a philosophical defence of something that should be decided by evidence. We find it surprising that Mohindra believes that logic should play no part in assessments of evidence; this brief reply analyses this and four other fundamental problems with his paper. In addition to this first error, Mohindra is mistaken about the evidence being in equipoise in his example case; his conclusion about cost-effectiveness does not follow; he is wrong to suggest that NICE is engaged in a de facto clinical trial; and is in error to suggest that anyone could ever win compensation because they were not given prophylaxis.

  Philosophy and logic

  We argued in our paper that NICE's decision was logical and sensible, given the lack of evidence for any effectiveness of antibiotic prophylaxis (AP) in preventing infective endocarditis (IE), and that those who disagreed were putting faith before logic and evidence.[1] In explaining his reaction to our paper, Mohindra states that "The source of my surprise was the fact that both philosophical and theological argument had been brought to bear in order to persuade clinicians of the merits of this particular piece of NICE guidance. Guidance that should have been driven by the empirical data."[2] Here Mohindra suggests by implication that the NICE guidance itself was based on "philosophical and theological argument" rather than evidence, which does not follow from the fact that our paper dealt in part with such arguments. More importantly, our paper did not use theological argument: it pointed out that the attitudes towards evidence displayed by many cardiologists are similar to those of some religious people. If anything, our argument is more evidence-based than his own: we criticised some cardiologists for faith in the face of evidence, and his paper seems to be a further example of this. In addition, despite criticising our paper, he offers no rebuttal of any of its key arguments. It is also somewhat ironic that Mohindra criticises our use of philosophy rather than evidence in a clinical journal when he claims to regard logic and evidence as the crux of the matter. It shouldn't need saying, but logic is philosophy, and the scientific method is a type of applied philosophy. Criticising someone for using philosophical argument is the same as criticising him for using logical argument.

  Evidence and equipoise

  Mohindra is certainly right to point out that there is not much evidence regarding the benefits and harms of AP. He quotes the NICE statement that "There is insufficient evidence to determine whether or not antibiotic prophylaxis in those at risk of developing infective endocarditis reduces the incidence of IE when given before a defined interventional procedure (both dental and non-dental)."[3] From this Mohindra concludes that "a fact neither formally proved nor formally disproved lies at the heart of the NICE guidance." He goes on to argue that this is different from a normal case of equipoise, where there is approximately equal evidence both for and against an intervention; in the case of AP for IE, there is virtually no evidence for or against, so this is a case of what he calls "insufficient evidence clinical equipoise" rather than the usual "competing evidence equipoise":

  "There is evidence that carries adequate probative force both for and against the proposition (competing evidence equipoise); or there is insufficient evidence to support the determination of either the truth or falsehood of the proposition (insufficient evidence equipoise)."

  Unfortunately, this appears to be an attempt to create a false distinction. What does "adequate" mean here? If the evidence both proves and disproves the proposition, then there is insufficient evidence to support the determination of either the truth or falsehood of the proposition; the two supposedly different types of equipoise are actually identical. Equipoise is simply the state of being in equilibrium, and clinical equipoise is the term we use for the evidence base being in equilibrium. This is true whether there is no evidence on either side or a great deal, but the same amount, on each side. In fact, "insufficient" in this context means that there is not enough evidence to decide one way or the other, so cases of "competing evidence equipoise" are the same because there is insufficient evidence on either side to make a decision.

  Cost-effectiveness and resistance

  Mohindra argues that we can make a decision based on a value judgement in cases of "competing evidence equipoise" but that to do so in cases of "insufficient evidence equipoise" is to make a value judgement that by necessity cannot be based on any evidence of cost-effectiveness, as there is no evidence concerning effectiveness. He does not explain why the same is not true in cases of competing evidence equipoise, which again suggests that this is a false distinction. Nonetheless, he states that:

  "Two points emerge from this position. First in the absence of adequate empirical evidence the use of other expert opinion as a foundation for new expert opinion based guidance does not move the evidential ball at all. Second, and critically, NICE cannot logically reach a conclusion upon cost-effectiveness in the absence of adequate evidence of effectiveness or ineffectiveness."

  Regarding the first point, NICE never claimed to have produced new evidence. They simply put an end to the practice of using an intervention for which there is no evidence base. It is somewhat ironic that Mohindra should criticise NICE for basing their guidance on expert opinion and producing no new evidence when he seems to favour continuing to give patients a treatment for which there is no evidence because of the faith- based expert opinion of some cardiologists. With regard to the second point, it is indeed true that a typical decision about cost-effectiveness cannot be made in the absence of evidence of effectiveness. But in the absence of evidence of effectiveness, there is no reason whatsoever to even consider cost-effectiveness; it would be irresponsible to spend money on treatment that is not supported by evidence. Furthermore, as we pointed out in our original article, there is also the cost of increased resistance to antibiotics and consequent harm to patients, which Mohindra fails to consider.[4] If the evidence regarding benefit to the patient is in equipoise, but the treatment costs money and will contribute towards the development of antibiotic resistance, the overall picture is that this is not a case of equipoise at all.

  Trials and tribulations

  Mohindra argues that NICE should have concluded that more research is needed into the use of AP to prevent IE. He also claims that NICE is essentially running an unofficial trial: "the effect of the NICE guidance is to institute a de facto population wide prospective cohort study to test the null hypothesis that antibiotic prophylaxis does not reduce the population wide incidence of IE". Is this true?

  It is clearly not true. If anyone is guilty of conducting an unethical trial, it is cardiologists who gave AP in the past or continue to do so. As stated above and in our original article, NICE concluded that AP should not be prescribed because there is no evidence for its efficacy. Mohindra himself suggests that more research is needed because the evidence is in equipoise. This means that a trial is necessary in order to justify the use of AP for IE. Clinical trials test new interventions for which there is no evidence base. But if the evidence is in equipoise, it is giving prophylaxis that is the trial, not withholding it. If NICE had allowed the widespread use of AP to continue, that would have been a clinical trial by stealth, and that was what was happening before the new guidance was issued: patients were being used as trial subjects because the collective wisdom of cardiologists was that AP prevented IE. Such faith-based experimentation in the face of the evidence is exactly what our original paper targeted, and NICE actually put an end to "a de facto clinical trial protocol outside the pre-existing legal regulatory framework", as Mohindra calls it.

  Compensation and consolation

  Finally, Mohindra's point about compensation is deeply flawed. He references Connaughton, who has suggested that "the NHS might adequately compensate anyone whose health suffered as a consequence of the new guidelines."[5] This whimsical wish could never be fulfilled. It would be impossible to prove that an incidence of IE was caused by a failure to administer AP in terms of causality, but also because there is no evidence that AP prevents IE anyway. This argument would only work if there was evidence of IE's efficacy, and Mohindra's use of the argument suggests he has faith that AP does work despite the lack of evidence; in other words, he is committing the very error that we highlighted in our original paper. It is also irresponsible to suggest that patients could claim compensation when they have no chance of doing so. Contrary to Mohindra's claim, this view has no legal or ethical force.

  Conclusion

  We argued in our original paper that some cardiologists resemble the people who accepted Pascal's Wager [6]. Just as someone might choose to believe in God because of the prospective benefits and despite the lack of evidence of His existence, so some cardiologists choose to believe in the efficacy of AP, despite the lack of evidence that it works, in a modern example of the so-called "no-lose" philosophy in medicine.[7] In his attempt to discredit NICE's guidance on IE, Mohindra commits several logical errors, and shows that his beliefs are based on faith rather than evidence. Ironically, he argues that "absence of evidence does not imply evidence of absence" to make the point that the lack of evidence does not mean that AP is not effective against IE, seemingly unaware that this very statement is frequently used as a theological argument for the existence of God, just like Pascal's Wager. We seem to have come full circle. Both logic and evidence indicate that NICE's guidance is quite correct. Mohindra's suggestion that NICE is engaged in an illegal trial is irresponsible, as is his suggestion that compensation could be claimed for harm caused by omitting AP. Despite Mohindra's attempt to introduce a false distinction, equipoise is equipoise, and there is not equilibrium in this case.

  References

  1. Shaw D, Conway DI. Pascal's Wager, infective endocarditis and the "no-lose" philosophy in medicine. Heart 2010;96:15-18.
  2. Mohindra RK. A case of insufficient evidence equipoise: the NICE guidance on antibiotic prophylaxis for the prevention of infective endocarditis. J Med Ethics doi:10.1136/jme.2010.036848 (online early).
  3. National Institute for Health and Clinical Excellence. Clinical Guideline 64: prophylaxis against infective endocarditis. http://www.nice.org.uk/nicemedia/pdf/CG64NICEguidance.pdf (accessed 9 Aug 2010)
  4. Lewis MA. Why we must reduce dental prescription of antibiotics: European Union Antibiotic Awareness Day. Br Dent J 2008;205:537-8.
  5. Connaughton M. Commentary: Controversies in NICE guidance on infective endocarditis. BMJ 2008;336:771.
  6. Pascal B. Pensees (no 233). http://www.gutenberg.org/files/18269/18269-h/18269-h.htm (accessed 9 Aug 2010)
  7. Galbraith S. The "no lose" philosophy in medicine. J Med Ethics 1978;4:61-3.

  Conflict of Interest:

  None declared

  Show Less

  Conflict of Interest:
  None declared.

  • Back to top