Contributed Cornrnents*

The advocacy of a causal relationshipbetween the smoking of tobacco and pulmonary cancer has developed some of theprodromata of an apostolic crusade,though virtually the entire basis for thisbelief is a statistical association. In a published review (1955) of the HammondHorn report (1954) Berkson, the distinguished biometrician at the Mayo Clinic,remarked that he is not affected by “¿�theconsiderable number of statistical studiesthat have been published . . . showing anassociation between smoking and cancerof the lungs. On the contrary. undeviatingconsistency of statistical results all in support of the same conclusion is in some circumstances the hallmark of spurious statistical correlation.― In 1957 Dr. Berkson,in a communication to a Congressionalsubcommittee, reiterated his earlier statement: “¿�Since(1955) more data have appeared, which I have studied carefully.I find no reason to modify my previousopinion that the evidence, taken as awhole, does not establish, on any reasonable scientific basis, that cigarette smoking causes lung cancer. On the contrary, Ihave found more and stronger reasons todoubt this conclusion.― Some of the morenotable discrepancies can be listed withnecessary brevity:

1. The “¿�epidemic―increase in lung cancer may be more apparent than real, anddue in great part to a wider availability ofimproved diagnosticskills.

2. While theproportionofwomen smoking has increased from less than 5 per centto 40 per cent,lung cancerhas emergedas a disease of males from an earlier ratioof 2:1 to about6:1.

3. The median age at diagnosis has notchanged.4. Based on mass samples, correlation islacking.

‘¿�Opinionsexpressed are not necessarily those ofthe A nnerican Cancer Societ;'.

RelativeCigarette

Consumption

RelativeLung Cancer

Mortality

1:2U.S.:England 2:1

NewYork:ldaho 1:1 4:1

Charlotte(N. C.): 1:1 1:4Manhattan

5. The urban:rural ratio of lung cancerfor nonsmokers, by the British EmpireCancer Campaign Report for 1956,was9:1, or roughly the observed difference inatmospheric pollutants.

6. There is a distinctrelationshipbetween socio-economic status and pulmonary cancer, with no comparable disparityin smoking habits.

7. Experimentalpulmonary carcinomaisreadily evoked by an environment of artificial “¿�smog―(Kotin and associates),while similar exposure of laboratory animals to cigarette smoke has yet to produceany significant neoplasia of the respiratorytract.

The most disturbing feature of prematurely designating tobacco as an established causative agent is an inevitableobscuration of the problems inherent inthe “¿�atmosphericsewers― of our urbanareas, with a predicted increase of 30-fold in synthetic chemical manufacturealonein the next 20 years.The averageurbanite inhales some 8,000 quarts of hissullied atmosphere each 24 hours. Somedilution of this noxious ambient by theambrosial products of the solacing weedmust remain as an inalienable, individualprivilege. The etiologic possibilities ofgenetic, hormonal, and environmental factors in lung cancer are so complex that theproposal for anti-tobacco propaganda onsuch a basis amounts to intellectual excess.

Ian Macdonald,M.D.Assoc. Cli,z. Professor of Surgery, Univ. of South

ern California School of Medicine, Los A ngeles.

70

I. Chinks in the Statistical Armor