Is ‘But-For’ The Best Tool For Applying Causation

Is ‘But-For’ The Best Tool For Applying Causation


The causal link is a crucial element of a case where the fault of the defendant, with regard to the injury caused to the claimant, needs to the proved. The ‘but for’ test, one of the forms of causation and also known as factual causation, is used to establish a causal link between the tort suffered by the claimant through the actions of the defendant. This article will look into how this test for factual causation (‘but for’ test) is used to link responsibility to the defendant. The exceptions to this test including strict liability cases and other forms of causation that might influence the common understanding of the nature of cause i.e. the attribution of responsibility or blame. An analysis of cases will be presented showing how courts are moving away from the standard use of the ‘but for’ test of causation.

Factual causation

The ‘but for’ causation is a test used by the court to establish fault of the defendant which caused damage to the claimant. This test would in turn help determine what the position of the claimant would have been had it not been for the defendant’s breach of duty. In other words this is the assessment of the responsibility that the court believes should be associated with the defendant for the claim in question.

The purpose of a claim in tort is to put the defendant back in the position where he or she might have been had the tort not occurred. Therefore, the ‘but for’ test is used to establish the effect that the defendant’s acts (and in some cases omissions) had on the situation of the claimant. The rationale for the critical importance of the ‘but for’ test is contained in the elegantly simple remarks of Major J in a Canadian case [1] where the emphasis of placing the claimant in the original position is stressed by stating that it is not the aim in tort law to position the claimant in a position which is better than the original one. The judgment states:

“The essential purpose and most basic principle of tort law is that the plaintiff must be placed in the position he or she would have been in absence of the defendant’s negligence. However, the plaintiff is not to be placed in a position better than his or her original one. It is therefore necessary not only to determine the plaintiff’s position after the tort but also to assess what the “original position” would have been. It is the difference between these positions, the ‘original position’ and the ‘injured position’, which is the plaintiff’s loss.”[2]

‘But for’ test to attach blame

The essence of the ‘but for’ test is explained concisely in the case of Cork v Kirby MacLean [3] where an epileptic workman fell to his death after a fit because of the absence of guard rails. The employers claimed that they did not cause his death rather it was caused by epilepsy. Lord Denning in the judgment said:

“If the damage would not have happened but for a particular fault then the fault is the cause of the damage; if it would have happened just the same, fault or no fault, the fault is not the cause of the damage.”[4]

These comments show how the courts need to use this factual causation test to go to the very roots of the fault, where responsibility is assigned to a party if they in fact caused the damage. This same definition leads us to the question about the application of this criteria in case of an omission.

The statement by Lord Denning explaining the ‘but for’ test seems very straightforward when applied to an act. However, consider the issues that might arise in cases dealing with a harm caused by an omission. Can the question then be ‘would the damage have occurred if the defendant had acted?’

Is the ‘but for’ test unsuitable for omissions?

In the Canadian article Proving Causation Where The But-For Test Is Unworkable [5], the question put forward is whether the traditional use of the ‘but for’ parameters is appropriate in situations where a failure to act caused the injury. The answer provided is that it is unsuitable. The more fitting question in such cases would be ‘whether the same injury would have occurred even if the defendant had acted’. This article is therefore putting forward an alternative approach to causation, moving away from the standard ‘but for’ test (reserving it for acts) to an ‘even if’ test for omissions. This might seem a very subtle change than the original test for factual causation but there might be serious issues by practically using them in court. Some might even be unable to identify the difference in the two questions. There might be situations where even an act which, according to the logic in the article[6], would normally use the ‘but for’ test, applies the test for omission, which would in practical terms waive off some or most of the liability. For example, if in a car accident where the claimant gets hits by the car of the defendant, the ‘but for’ question would be ‘would the claimant have been injured but for the defendant’s negligent driving?’ The answer to this could be no, which means that blame is attached to the defendant for his or her act of driving the car negligently causing the claimant to be injured. Keeping in mind that an objective answer is required, contrast this with the other question ‘would the claimant had been injured even if the defendant drove carefully?’ The answer to this can be yes, since there is always a risk involved on the road. Therefore the presence of these two questions for the establishment of causation would create significant injustice as well as uncertainties in the law. This is mainly because these can be manipulated to suit the specific case requirement which might not necessarily be just. This also signifies the limited scope of the ‘but for’ test, needing the development of some other criteria to assess the blame for omission. This could clearly be seen as creating uncertainties in the law itself.

The positive aspect about this could be developed by using both tests together to establish factual causation. This would mean that it would be much harder to prove a causal link. However, it would also mean that the blame which is ascertained at the end would be accurate and would have considered external factors affecting the situation in question as well.

Departure from the standard criteria of ‘but-for’

Lord Hoffmann in his article Causation [7] discusses how the but-for test is effective where the defendant was the sole cause of the injury with no other forms of interference [8]. The difficulty arises however, where it is inapplicable as we saw in the case of an omission. Lord Hoffmann presents another issue which was referred to in the introduction of this essay as an exception to the causation requirement for attribution of responsibility [9]. The whole idea about liability or blame in negligence is explained in the judgment of Kuwait Airways Corp v Iraqi Airways Co [10]: “The law never says simply that one is liable. One is always liable for something”[11]. This requires a clear establishment of a causal connection. Again this would be really straightforward in cases where a simple act of alleged negligence was involved and the original application of the causation test (‘but for’) was useable. However, problems arise as we have seen, in cases of omissions. Here departure from the ‘but for’ test is essential to reach any fair conclusion. This could be in the form of an alternative test to attribute responsibility to the defendant, or just widening the traditional ‘but for’ test.

Taking this problem one step further, we can assess situations such as those involving strict liability claims. This is a situation where, there is no need to establish the causal link hence no need to assign blame. An important case to illustrate this problem where strict liability rules apply is the case of Environment Agency v Empress Car Co (Abertillery) Ltd [12] where the attribution of responsibility for pollution of a river was in question. This case involves s.85(1) of the Water Resources Act 1991 [13] under which it is a criminal offence [14] to cause polluting matter to enter into a river. The courts construed it as a strict liability offence. For the Empress Car Co case, a tank of oil was placed near the river which was vandalised and resulted in the oil draining into the river. The court held that even though it was someone else’s act that caused the pollution, under the Water Resources Act it was considered under strict liability and it was enough to own a tank of oil to be responsible for the pollution. Therefore the owners of the oil tank could be penalised for pollution of the river under Section 85(1) of the Act. Considering this, some laws do not require any attribution of responsibility, hence in these situations the ‘but for’ test is the least important element in attaching responsibility. This, however, makes it easy on one hand to punish or fine someone as a certainty, while on the other hand makes room for an unfair and unjust system to develop. Lord Hoffman in his article [15] considers this as moving away from the standard criteria of causation.

Recent cases have appreciated the complications that might arise from the standard use of the ‘but for’ test. The decisions of some cases in the past decade or so have therefore seen the courts moving away from the traditional test of causation. The way this is done includes diluting the process which specifies the causal link between the injured party and the defendant. Tools such as material increase in risk and loss of chance have been integrated within the causation-determining process.

This approach has widened the scope of liability by making the establishment of cause easier by making the test for causal link more flexible.

In a way it is arguably important as well to move away from the standard use of the ‘but for’ test as in certain situations this could help in reaching a just decision. Disputes in cases such as those that involve for example multiple concurrent causes, multiple consecutive causes and loss of expectation are too complicated to be solved with the use of the simple factual causation test.

The situation where multiple factors are in play at the time the injury arises (known as multiple concurrent causes), it is close to impossible to use the ‘but for’ test as an absolute test of causation. Moving further away from the narrow test of factual causation, the court instead asks if negligence of the defendant has played a substantial role in materially increasing the risk of the injury. In the case of McGhee v National Coal Board [16], the court moved away from the standard test and decided to assess liability based on the question whether negligence was such that it materially increased the risk of damage. This was a bold step which made the factual causation test too wide, making it easier to prove liability based on partial blame.

This rule of material contribution however can be distinguished in cases involving an injury which is claimed for, or in the presence of multiple employers, where it is difficult to apportion responsibility of the damage suffered. This is especially important in cases such as asbestosis cases where the employee was exposed to asbestos during multiple employments. The blame here cannot be associated with one employer. In the case of Holtby v Brigham & Cowan (Hull) Ltd [17] even though the court did not apportion the blame on one employer, they did not have a set formula to allocate responsibility which is seen in the Court of Appeal decision of this case where the judges thought that a 50% reduction in damage would have been a suitable allocation of burden.

Causation rules however further widened two years later in the case of Fairchild v Glenhaven Funeral Services Ltd and others [18]. This case also involved an asbestos related disease but due to the presence of multiple employers, it was impossible to establish to a certainty, which of them caused it. The decision in this case was perhaps one which had the most impact on the historical use of the ‘but for’ test and how it would be used in the future. This decision was based on the requirement of proof on a balance of probabilities. This balance of probabilities was based on the Roman approach where the jurist could not tell who caused the harm, but they found it appropriate to make them all liable and then let them sort out the apportionment of liability among themselves [19]. So one employer can be sued who will then have the opportunity to claim shares in liability with other employers.

The most recent case and one that illustrates how far courts have moved from a standard use of the ‘but for’ test is the case of Chester v Afshar [20] in 2004. This is a non-disclosure case where surgery had a 1-2% risk of causing physical damage even if the surgery was performed without negligence. The surgeon failed to inform the claimant of the risk. The claimant argued that she would not have had the operation at that very occasion and so could have avoided the risk. The defendant argued that it was uncertain if she would have put off the surgery forever and the risk would still have been the same later. The defendant was still held liable and his appeal was dismissed.

This is a very peculiar situation as there was still no definitive association of blame with the surgeon (defendant). There was a greater probability that it would have happened regardless. The liability here was based on the omission to inform. The fact that the risk of injury was very low (1-2%), was considered to be unimportant. The causation question asked was “would the risk not have occurred when it did?” [21]

This can again be seen as a loose application of the causation test moving away from the usual ‘but for’ test. This case is complex because it involves an omission (to disclose information) as well as an act which caused the harm. The problem in the claim that caught the attention of most commentators was that had the claimant known the risk, she would not have had the procedure right away, but the risk still existed. The question here is whether it is justifiable to attach blame on speculation? This arguably is what the court avoided by holding the first surgeon responsible, given the uncertain nature of the facts (merely based on non-disclosure). The attribution of blame in such a case is merely speculative which carries with it the question of whether there is in fact any substantive value of the ‘but for’ test or whether it is just procedural.


It can be concluded that the use of causation is greatly dependent on the arguments presented with relation to negligence and stricter cause on one hand and damage on the other. This explains why causation does not have a shape of its own, which explains why it is so subjective even though the courts intended it to be objective. C V Dam mentions causation in her book as being “an elastic feature which can be stretched and shrunk according to the magnitude of other requirements”[22]. The fact that the burden of proof with regard to causation can be partially shifted from the claimant to the defendant in certain situations adds further to the flexible concept of the causation. Therefore the ‘but for’ test for causation also has some very obvious and some not so obvious shortcomings. The case law however is constantly evolving to make the tools used for attribution of responsibility as precisely applicable as possible, making sure the burden is justly attributed to the parties in question. Lord Hoffman, defending the rules of causation, stated in his article [23] that they were not intuitive but reasoned and the courts were capable of explaining the logic behind them. However he continued by saying that the law of causation still needs some clarification as to why it sometimes uses the standard criteria (‘but for’) and sometimes departs from them to establish causation and attribution of responsibility.



[1] Athey v. Leonati (1996) 140 D.L.R. 235

[2] Athey v. Leonati (1996) 140 D.L.R. 235 at para. 32

[3] Cork v Kirby MacLean [1952] 2 All ER 402

[4] Cork v Kirby MacLean [1952] 2 All ER 402, 407

[5] H David, W P,McCague and P F Yaniszewski, ‘Proving Causation Where The But For Test Is Unworkable’ [2003] TAQ 216, 222

[6] H David, W P,McCague and P F Yaniszewski, ‘Proving Causation Where The But For Test Is Unworkable’ [2003] TAQ 216

[7] L Hoffmann, ‘Causation’ [2005] LQR 592,

[8] L Hoffmann, ‘Causation’ [2005] LQR 592, 594

[9] L Hoffmann, ‘Causation’ [2005] LQR 592, 594

[10] Kuwait Airways Corp v Iraqi Airways Co (Nos 4 and 5) [2002] 2 WLR 1353

[11] Kuwait Airways Corp v Iraqi Airways Co (Nos 4 and 5) [2002] 2 WLR 1353, para   128 Lord Hoffman

[12] Environment Agency v Empress Car Co (Abertillery) Ltd [1998] 2 W.L.R. 350

[13] Water Resources Act 1991 s 85(1)

[14] it is noted that this is a criminal case but for the purposes of our essay the attribution of responsibility is what needs to be considered

[15] L Hoffmann, ‘Causation’ [2005] LQR 592

[16] McGhee v National Coal Board [1973] 3 All ER 1008

[17] Holtby v Brigham & Cowan (Hull) Ltd [2000] 3 All ER 421

[18] Fairchild v Glenhaven Funeral Services Ltd and others [2002] 3 W.L.R. 89

[19] L Hoffmann, ‘Causation’ [2005] LQR 592, 599

[20] Chester v Afshar [2004] 3 W.L.R. 927

[21] T Clark and D Nolan, ‘A Critique of Chester v Afshar’ [2014] OJLS 659, 662

[22] C V Dam, European Tort Law (2nd, OUP, Oxford 2013) 309

[23] L Hoffmann, ‘Causation’ [2005] LQR 592, 602-603

The views expressed in this article are those of the author and do not necessarily represent the views of or any organization with which he might be associated.

Muhammad Mabroor Jameel

Author: Muhammad Mabroor Jameel

The writer holds an LLB (Hons.) degree in Maritime Law from the University of Southampton. He has also worked in Mayfair London.