Occupational and Environmental Medicine 1995;52:289-293

Cause of occupational disease

D C F Muir

AbstractThe concept of causality is reviewed withspecial emphasis on occupational dis-eases. Separate approaches from thephilosophical, scientific, and legal pointsof view are identified. There is contro-versy over the methodology of logicalcausality; inductive and deductive meth-ods are described and reference is madeto the verification or refutationapproach. Application of the methods toepidemiology are reviewed.

It is likely that many diseases havemultiple causes and that only a compo-

nent of occupational causality can beidentified in each patient. Methods ofassigning such a component are dis-cussed. The difficulties of developing an

equitable compensation policy in suchcircumstances are reviewed. The possiblebenefits of proportional compensationare noted.

(Occup Environ Med 1995;52:289-293)

Keywords: causes; compensation; occupationaldiseases

Occupational HealthProgram, McMasterUniversity, Faculty ofHealth Sciences,Hamilton, Ontario,CanadaD C F MuirCorrespondence to:Dr D C F Muir,Occupational HealthProgram, McMasterUniversity, Faculty ofHealth Sciences, 1200 MainStreet West, Hamilton,Ontario L8N 3Z5, Canada.

Accepted 19 January 1995

Identifying the cause of disease in a popula-tion or in an individual person may be diffi-cult and the criteria for recognition ofcausality have been considered by severalauthors. Causality is particularly important inthe case of occupational diseases because pre-vention is usually more important than thera-peutic intervention and also because many

countries have developed policies for thefinancial support of victims of occupationaldiseases that are more generous than thoseprovided for other disabled people. The pur-pose of this article is to review the concept ofcausality as it was developed in classicalphilosophy, developments in epidemiologicalthought, and the impact of legal analysis. Theemphasis is on causality of occupationaldiseases and related policy issues.

The nature of causalityThe philosophers of early civilisationsobserved the natural world and proposed vari-ous explanations for what they saw. FrancisBacon in 1599 considered these to be super-stitions that should be distinguished from the

methodical approach of the true scientist.1The true scientist in his view, collects empiricalobservations in an unbiased way and, -with an

open mind, infers the cause of events from theevidence. The terms inductive and deductiveinference are exemplified as follows. Veryextensive epidemiological surveys on thehealth of cigarette smokers have been carriedout. The observations induce the inferencethat smoking is a major cause of lung cancer.This argument in inductive logic is from par-ticular examples to a general rule. On theother hand, if it is once accepted as a generallaw or rule that cigarette smoking is a cause oflung cancer, then a physician may argue fromthe general to the particular and make thedeductive (or hypotheticodeductive) inferencethat smoking was the most important cause oflung cancer in his particular patient. Thisexample illustrates deductive logic from thegeneral to the particular.The Scottish philosopher Hume in 1739

questioned the validity of inductive logic fromrepeated observation to general laws ofcausality.2 In a formal logical analysis, heclaimed that repeated and consistent observa-tions of two associated events could not beused as evidence that one caused the other.This was the start of a major philosophicaldispute. One school (the positivists or verifica-tionists) claim that sufficient and increasingnumbers of observations do provide a persua-sive body of evidence of causality that workson a practical day to day basis. A certainframework must be used to avoid nonsensicalassociations. With that reservation, the posi-tivists maintain that the inductive methodoffers sensible approaches to public healthpolicy.An opposite approach was favoured by Karl

Popper who was directly opposed to allattempts to operate with the idea of inductivelogic.3 in his view, simply collecting furtherobservations that confirmed an associationbetween two events could not contribute inany way to the debate as to whether onecaused the other. He suggested that anexplanatory theory connecting two events wasfirst developed or invented in the mind of acreative scientist. Certain deductions or pre-dictions could be made from the theory andthese could be tested. If the conclusionsturned out to be acceptable then the theoryheld for the time being, but this did not estab-lish the truth of the theory or even its probabletruth. If the conclusions were falsified, how-ever, then the theory itself must be false. Theapplication of Popper's ideas to human epi-demiology was first proposed by Buck.4 Thiswas followed by a vigorous debate in whichother epidemiologists5 emphasised the value,

289 on June 28, 2020 by guest. P

rotected by copyright.http://oem

.bmj.com

/O

ccup Environ M

ed: first published as 10.1136/oem.52.5.289 on 1 M

ay 1995. Dow

nloaded from

290

from a practical and policy point of view, ofcareful and repeated observations on definedpopulations with reasoned inductive infer-ences as to causality.

Criteria for causality in epidemiology wereproposed by Hammond in 1955,6 Yerushalmyand Palmer in 1959,7 and in the landmarkReport of the Surgeon General's AdvisoryCommittee on Smoking and Health in 1964.8A framework for drawing inductive conclu-sions was formulated and was used to identifycigarette smoking as the major cause of lungcancer. The framework was developed moreconcisely by Bradford Hill whose name isfirmly associated with inferences of causality.9His suggestions for considering causality arewidely taught and are as follows:the strength of associationconsistency specificitythe relation in timethe biological gradientbiological plausibilitycoherence of the evidencethe experimental evidencereasoning by analogy.

Hill wrote that none of these nine view-points could bring indisputable evidence foror against a cause and effect hypothesis andequally none could be required as essential (asine qua non). What they could do, withgreater or less strength, was to answer the fun-damental question-is there any other way ofexplaining the set of facts? Was there anyother answer that was more likely than causeand effect?

In a formal sense Hill's proposals are sub-ject to the same underlying weakness ofinductive logic described by Hume so manyyears ago. Unease about the lack of distinc-tion between association and causation in thecase of smoking and lung cancer wasexpressed by such a formidable statistician asR A Fisher.'0 In a series of essays, supportersof Bradford Hill's inductive method or, con-versely, of Popper's deductive refutation,presented their arguments and counterarguments." For the general reader it is ofinterest to note that opinion is divided.Neither Bradford Hill's nor Popper's methodreceive universal support. The idea of usingthe Bradford Hill criteria in a Popperian sensewas suggested by MacClure"2 whereby acausal hypothesis based on inductive logicgenerates deductive predictions that couldthen be tested. If the Bradford Hill criteriafailed to reject the hypothesis, then it couldcontinue to be used for making day to daypractical decisions, although fulfilment of thecriteria did not, of itself, constitute formalproof.The above analysis of causality refers essen-

tially to single causes of disease. This isbecoming uncommon in clinical medicineand the next section considers the conse-quences of multiple causality.

The nature ofmultiple causalityOccupational epidemiology examines therelation between disease in populations and

concomitant toxic exposures. Clinical occupa-tional medicine and the legal framework aremore concerned with attribution of causalityin the individual worker. It is increasinglyclear, however, that many illnesses are the cul-mination of multiple causes. In a populationof workers exposed to dust and to cigarettesmoke, certain people will develop chronicobstructive airway disease. In some, the majorcause of airway damage may be cigarettesmoking and in others, it may be dust expo-sure. The relative importance of each willdepend on the duration and concentration ofexposure to dust or to cigarette smoke. It isnot appropriate to treat either agent as if itwere the single effective cause.

Attribution can be achieved by applyingdata from group epidemiological studies tothe individual case. The increased risk of lungcancer in groups of workers who smoke 40cigarettes a day is known. If a patient smokesthat number, then it is not unreasonable tomake the assumption that the same risk isborn by that patient. When more than onecause is to be evaluated in the patient it is pos-sible to assign the proportion of harm fromeach source by the use of epidemiological dataincluding cohort or case control studies.'3This method has been used in deriving tablesof proportional risk of malignant diseasecaused by exposure to radiation.'4 Theapproach has been criticised on statisticalgrounds and the unmeasurable issue of indi-vidual susceptibility. No alternative method,however, seems to be available.When two or more factors cause a disease

the question may be raised as to whether oneor other component made a significant contri-bution. A physician can reasonably hazard aninformed guess whether some event such asan infection made any clinically significantdifference to the inevitable train of events in apatient with untreatable malignant disease.The same value judgement of significancecannot be used to evaluate the clinical impor-tance of a proportional estimate of causality.If cigarette smoking is estimated to accountfor 15% of the damage in a patient withobstructive airways disease, then this pointestimate cannot be further evaluated in termsof clinical significance.As physicians have struggled to analyse the

cause of disease in their patients a parallelbody of scholarship has been developedwithin the legal framework. To an extent theprofessions have remained separate and therehas been relatively little common exchange.The following paragraphs summarise the ideaof causality in law.

The legal frameworkAn extensive review of the development oflegal concepts of causation, starting from thedays of early Roman law was prepared byHonored (1986).' A distinction is drawnbetween scientific theories that attempt togive an account of the way in which eventshappen, and legal theories of causation andremoteness of damage that have the purpose

Muir on June 28, 2020 by guest. P

rotected by copyright.http://oem

.bmj.com

/O

ccup Environ M

ed: first published as 10.1136/oem.52.5.289 on 1 M

ay 1995. Dow

nloaded from

Cause of occupational disease

of defining how the limits of responsibilityshould be fixed. Some believe that roughcommon sense should suffice to define legalcausation but this is generally thought to betoo vague or imprecise. A difference in theprofessional approach is provided by theexample of the physician who is concernedwhether coal dust caused lung disease in aminer. A lawyer traces the sequence furtherback to see whether some wrongful act (fail-ure to guard against dust exposure) can beregarded as the cause. The lawyer provides afurther explanation of a different type.Honored traces the development of the philoso-phy of causation through Marx and Hegel andMill.'6 Although there are many publicationson the theory of legal causation Honore pro-poses that the ideas around which the deci-sions revolve are limited to necessity,explanation, probability, and legal policy.

NecessityThis has given rise to the equivalence theoryor essential condition (conditio sine qua non)theory. The crucial question is whether theevent was necessary for the production of theharm. Each condition is necessary to theproduction of harm and is in that sense equiv-alent (equally necessary). In most cases it isimpossible to apportion any particular part ofthe harm to any given condition. The rule,common in many countries, that independenttortfeasors, each of whom contributed to theharm, is wholely liable (in solidum) points tothe equivalence theory. The theory was morepopular in criminal cases and has fallen out ofuse in modern times.

ExplanationCauses of an event are those conditions thatexplain its occurrence. This extralegal idea isused in a negative way so that a tortfeasor isnot liable for harm that is adequatelyexplained by some other cause. This ideagenerates several theories. In the efficiencytheory, various conditions can be thought ofas making a quantifiable contribution to theharm. Those whose contributions come abovea certain percentage are causes of the harm,and the quantification can be used as a basisfor the apportionment of damage betweentortfeasors.

ProbabilityThe adequate theory-In its 19th centuryform, an essential condition (conditio sine quanon) is the adequate cause of harm if it is ofthe type that significantly increases the objec-tive probability of harm, of the type actuallysuffered. Harm can have two or more ade-quate causes, and they need not be of equalprobability of causing harm.

Legal policyThe theories are not strictly causal but refer tocodes of law, responsibility, and a sense ofequity between parties.The very brief summary of legal theory

shows clearly that the law is concerned withevaluating human conduct and not with inan-

imate objects. Whether asbestos causes canceris an issue of scientific fact. Whether theindustry that sells asbestos is involved in caus-ing cancer is an issue of legal causality. A keyquestion in decisions rests on the "but for"test. If harm would not have occurred "butfor" the defendant's actions, then thoseactions were a cause, not necessarily the onlycause of the harm.

The boundary between legal andscientific causationThe distinction between scientific and legalcausation or responsibility has beendescribed, but there are many situationswhere the two are almost inextricably linked.The most important is the problem of multi-ple causation and the questions of what con-stitutes a significant cause. The difficulty ofapplying legal concepts to scientific analysis isillustrated by the case of the asbestos workerwho smokes. Would he have developed lungcancer but for the exposure to asbestos?Obviously the answer can only be given inprobabilistic terms. The risk of developinglung cancer may have been much less in theabsence of asbestos, but was certainly notnegligible if he was a heavy smoker.Two much quoted judgements have influ-

enced legal thought. In Bonnington Castingsv Wardlaw (1956),'7 a worker developedsilicosis as a result of exposure to silica from apneumatic hammer at which he worked andalso silica from swing grinders at the sameworkplace. No dust extraction equipment wasknown or practicable for use with the hammerand the employers were not liable, in law, forthe consequences, the health effects on theworker being covered by the regulations gov-erning pneumoconiosis in the UnitedKingdom at that time. Although the swinggrinders were fitted with dust extractionequipment this was not kept free fromobstruction and, in this respect, the factoryowners were in breach of their statutory duty.The evidence suggested that most of the dustcame from the hammer but the legal argu-ment centred on whether the dust from thegrinders could have been the cause of the dis-ease. In the House of Lords, judgement wasgiven that the cause of the disease could havebeen from either source. No attempt wasmade to take account of contributions fromtwo separate sources. It was held that theplaintiff did not have to show that the dustfrom the swing grinder was the sole or eventhe most substantial cause of his disease, if hecould show, on balance of probabilities, thatdust from that source had materially con-tributed to the disease. Anything that did notfall within the negligible (de minimum) princi-ple would add a material contribution. Theplaintiff won the case. It is evident that this istantamount to treating the pulmonary diseaseas a single indivisible harm. Scientifically,however, the degree of lung damage in silicosisis closely related to the amount of dust expo-sure and, when two sources can be identified,then the corresponding degree of harm can be

291 on June 28, 2020 by guest. P

rotected by copyright.http://oem

.bmj.com

/O

ccup Environ M

ed: first published as 10.1136/oem.52.5.289 on 1 M

ay 1995. Dow

nloaded from

Muir

apportioned. The worker might indeed havedeveloped a degree of silicosis from exposureonly to dust from the swing grinder but theamount of disability is likely to have beenmuch less. A further issue is that terms suchas material or significant contributions haveno defined medical meaning. Courts seem tobe willing to infer that a given cause makes amaterial contribution. This may respond toquestions of legal policy but there seems to beno valid scientific support for the proposition.A somewhat different situation occurred in

McGhee v the National Coal Board.'8 Aworker developed dermatitis due to dust fromtwo sources. One was an unavoidable compo-nent of his employment in a brick works andwas encompassed by appropriate legislation.The other was due to dust remaining on hisclothing as he cycled home, and was held tobe the responsibility of the employer whoshould have provided adequate shower facili-ties. The physicians who gave expert testi-mony did not seem to be able to quantify theharm from the two sources of dust and thedamage could not be apportioned. In theabsence of any such evaluation it was evidentthat the whole cause of the dermatitis could,in principal, have been due to dust on theworkers clothing and the plaintiff won. Thiscase is better known because of the nature ofthe legal tests of proof that were involved, butthis will not be discussed here.

Causation and public policyThe early literature contains many reportsthat concluded that there was insufficientproof that asbestos exposure caused cancer oreven that smoking caused cancer. Suggestiveevidence, but that more research is needed,seemed to be the view of many writers.Perhaps lack of evidence was not the realissue. The ill effects of smoking are no longerseriously disputed, but the overall world salesof tobacco are rising. The certainty of causalityrequired in making public policy depends onthe potential for harm and the number ofpeople affected. Risks and benefits must beweighed. Rothman and Poole point out thatpractical policy making is really a politicalprocess and, in their view scientists should notallow themselves to be drawn into this duringworking hours.'9 Their estimates and interpre-tations of causality should be as unbiased as ispossible. If a given agent is thought to be apossible cause of human disease on weak evi-dence, it may be reasonable to prevent itbeing used in the first place. Removing anagent, which has been present for many years,might require much more evidence of causalitybefore huge costs are incurred.

Causation and toxic tort litigationThe cause of occupational or environmentaldisease becomes an issue to be decided, notby scientists, but by a non-medical jury in themany cases that have come to be called toxictort litigation in the United States. In areview, Foster et al note that the court mustassess the evidence and also the credibility ofexpert witnesses who present the evidence.

The doubtful reliability of some of the expertwitnesses has created what one author hascalled junk science.2' As far as the evidenceitself is concerned the plaintiff in these casesmust prove that the exposure, more likelythan not, caused the injury. Admissible evi-dence, based originally on the Frye (1923)22rule now accepts federal rules of evidence,23which, among other criteria, evaluates evi-dence of causation after taking account of thetestability of the theory, peer review, andsources of potential error.

Significant references exist in the legal liter-ature on the use of epidemiological evidencefor attribution of causality.'4 Some think thatcomplex issues of risk assessment should notbe left to trial lawyers, judges, and juries.25

DiscussionThe central theme of this review is that identi-fication of the cause of diseases related tooccupation is difficult. Unambiguous proof israrely obtained and is not demanded by physi-cians familiar with statistical and epidemio-logical concepts. The nature of causality iscomplex and the scientific and legalapproaches can be quite different. Many dis-eases, both occupationally related and other-wise, have multiple contributing causes. ThePearson Commission thought that multiplecausality would be an increasing problem inthe future.'6 Causality has a very special placein occupationally related diseases because ofthe issues of prevention and compensation. Asfar as prevention is concerned it does not mat-ter whether lung cancer in asbestos workers isdue both to smoking and to asbestos.Reduction or removal of both is important.Compensation is more difficult, however,because of the practice of awarding full com-pensation even though only a part of the causein an individual person is due to occupationalfactors. The percentage point at which anoccupational component is considered suffi-cient to result in full compensation variesremarkably with a range from 50% in somejurisdictions to 5% in others. This has sub-stantial economic effects on the individualpeople and communities concerned. What-ever the cut off point it leaves certain workerswithout financial consideration even though aportion of their illness or disability was occu-pational in origin. Other workers may be over-compensated in the sense that only a portionof their illness or disability was due to occupa-tion. The origin of the difficulty is the imposi-tion of a binary decision process (fullcompensation or no compensation) on a con-tinuous distribution of causality. One solutionmight be to compensate on a sliding scale inproportion to the amount of occupationalcausality.'7 If disease in a worker, whetherobstructive airways disease, lung cancer, orany other condition can be estimated to havean occupational component of causality of(say) 30% then that is the proportion of thedisability that is compensated. Whatever theadministrative difficulties of such an approachit has the merit of consistency and logic.

292 on June 28, 2020 by guest. P

rotected by copyright.http://oem

.bmj.com

/O

ccup Environ M

ed: first published as 10.1136/oem.52.5.289 on 1 M

ay 1995. Dow

nloaded from

Cause of occupational disease

A more radical criticism of occupationalcompensation is that financial support to adisabled patient depends on causation ratherthan on need.28 This generates a privilegedclass of disabled within the community29 thatis the more illogical when occupationalcausality is only part of the origin of disabilityin an individual subject.

1 Bacon F, 1599. In: FH Anderson, ed. The New Organonand related writings. New York: Bobbs-Merrill, reprinted1960.

2 Hume D. Treatise of human nature. London: John Noon,1739. In: Selby-Biggs LA, ed. Revised and reprinted.Oxford: Clarendon Press, 1985.

3 Popper KR. The logic of scientific discovery. Revised ed. NewYork: Harper and Row, 1968. Originally published as:Logik der Forschung. Vienna: Springer, 1934.

4 Buck C. Poppers' philosophy for epidemiologists. Am JEpidemiol 1975;4: 159-68.

5 Jacobsen M. Against Popperized epidemiology. Int JfEpidemiol 1976;5:9-1 1.

6 Hammond EC. Cause and effect. In: Wynder ES, ed. Thebiologic effects of tobacco. Boston: Little, Brown1955:171-96.

7 Yerushalmy J, Palmer CE. On the methodology of investi-gations of etiologic factors in chronic diseases. J ChronDis 1959;1O:27-40.

8 Surgeon General's Advisory Committee on Smoking andHealth. Rockville, Maryland: Public Health Services;DHEW, 1964. (PHS No 1103.)

9 Hill AB. Principles of medical statistics, 9th ed. London:Lancet 1971:313-23.

10 Fisher RA. Lung cancer and cigarettes. Nature1958;132: 108.

11 Rothman KJ, ed. Causal inference. Chestnut Hill, MA:Epidemiology Resources, 1988.

12 MacClure M. Popperian refutation in epidemiology. Am JEpidemiol 1985;121:343-50.

13 Enterline PE. Attributability in the face of uncertainty.Chest 1980;78:377-9.

14 Lagakos SW, Mosteller F. Assigned shares in compensa-tion for radiation-related cancers. Risk Analysis1986;6:345-57.

15 Honor AM. In: Tunc A, ed. Causation and Remotenessof Damage. International encyclopedia of comparative law.Vol XI Torts, Chapter 7. North Virginia: Kluwer Ac NV,1986:71.

16 Mill JS. A system of logic, ratiocinative and inductive.London: Parker, Son, and Brown, 1872.

17 Wardlaw v Bonnington Castings Ltd. London: House ofLords, 1956 AC613 L(HL).

18 McGhee v National Coal Board. London: 1972; 3 All ER:108.

19 Rothman KJ, Poole C. Science and policy making. Am JPublic Health 1985;75:340-1.

20 Foster KR, Bernstein DE, Huber PW. Science and thetoxic tort. Science 1993;261:1509 and 1614.

21 Huber PW. Galileo's revenge: junk science in the courtroom.New York: Basic, 1991.

22 Frye v United States. District of Columbia: 1923. (293F1013).

23 Daubert v Merrill. Dow Pharmaceutical Inc. 1993;61USLW:4805.

24 Black B, Lilienfeld DE. Epidemiological proof in toxic tortlitigation. Fordham Legal Reviews 1984;52:732.

25 Elliott ED. Why courts? Comment on Robinson. Journal ofLegal Studies 1985 XIV:799-805.

26 Royal Commission on Civil Liability and Compensation.London: HMSO, 1978.

27 Muir DCF. Compensating occupational diseases: amedical and legal dilemma. Can Med Assoc J 1993;148:1903-5.

28 Ison TG. Compensation for industrial diseases under theWorkers' Compensation Act of Ontario. Industrial DiseasesStandards Panel, Government of Ontario.

29 Stapleton J. Disease and the compensation debate. Oxford:Clarendon Press, 1986.

Vancouver styleAll manuscripts submitted to Occup EnvironMed should conform to the uniformrequirements for manuscripts submitted tobiomedical journals (known as theVancouver style.)

Occup Environ Med, together with manyother international biomedical journals, hasagreed to accept articles prepared in accor-dance with the Vancouver style. The style(described in full in the BMJ, 24 February1979, p 532) is intended to standardiserequirements for authors.

References should be numbered consec-

utively in the order in which they are firstmentioned in the text by Arabic numeralsabove the line on each occasion the refer-ence is cited (Manson' confirmed otherreports2-5 . ). In future references topapers submitted to Occup Environ Med

should include: the names of all authors ifthere are seven or less or, if there are more,the first six followed by et al; the title ofjournal articles or book chapters; the titlesof journals abbreviated according to thestyle ofIndex Medicus; and the first and finalpage numbers of the article or chapter.Titles not in Index Medicus should be givenin full.

Examples of common forms of refer-ences are:

1 International Steering Committee of Medical Editors,Uniform requirements for manuscripts submitted tobiomedical journals. BrMedJ7 1979;1:532-5.

2 Soter NA, Wasserman SI, Austen KF. Cold urticaria:release into the circulation of histamine and eosino-phil chemotactic factor of anaphylaxis during coldchallenge. N EnglJ_7Med 1976;294:687-90.

3 Weinstein L, Swartz MN. Pathogenic properties ofinvading micro-organisms. In: Sodeman WA Jr,Sodeman WA, eds. Pathologic physiology, mechanismsof disease. Philadelphia: W B Saunders, 1974:457-72.

293 on June 28, 2020 by guest. P

rotected by copyright.http://oem

.bmj.com

/O

ccup Environ M

ed: first published as 10.1136/oem.52.5.289 on 1 M

ay 1995. Dow

nloaded from