Author Affiliation: Health Outcomes Research Group, Memorial Sloan-Kettering Cancer Center, New York, New York.
In this case-control study, the investigators compared the smoking histories of individuals with lung cancer and without lung cancer. Participants were matched for age and several dimensions of smoking history were ascertained through a standardized set of questionnaires (including age at initiation and cessation, and average amount smoked per day of cigarettes, cigars, or with a pipe). Each individual's occupational history and history of prior lung disease was captured to control for possible confounders. The investigators found that individuals with lung cancer had a more extensive smoking history than individuals without lung cancer.
See PDF for full text of the original JAMA article.
In 1950, when Wynder and Graham1 published their landmark article in JAMA about the link between smoking and lung cancer, the lung cancer mortality rate for men in the United States had already experienced a more than 4-fold increase over the prior 3 decades, and appeared likely to surpass cancers of the colon and rectum as the leading cause of cancer death among men.2 The increase paralleled the rise in cigarette consumption among men that had occurred roughly 20 years earlier.
A total of 684 patients with lung cancer were compared with several sets of control participants, including patients admitted to the hospital with non–lung cancer diagnoses. The authors found that the patients with lung cancer had a more prolonged and intense history of smoking than the control participants—a finding consistent with the hypothesis that smoking caused lung cancer. The investigators concluded, “From the evidence presented. . . the temptation is strong to incriminate excessive smoking, and in particular cigarette smoking, over a long period as at least one important factor in the striking increase in bronchiogenic carcinoma.”
The study by Wynder and Graham1 was published at a time before the causal link between smoking and lung cancer was an accepted scientific fact. Even after publication of the study1 and a later longitudinal analysis by Doll and Hill3 with similar findings, the link between smoking and lung cancer remained underappreciated and controversial.4
As evidence of the mindset at the time, an article published in Reader's Digest in 1952, provocatively titled “Cancer by the Carton,” described the debate over the role of cigarettes in causing lung cancer as “a medical controversy. . . largely kept from public notice.”5 Because Reader's Digest had a large circulation at the time, this article was 1 of the milestones in the history of public awareness of the topic, but the scientific community took more time to convince. Seven years after the study by Wynder and Graham1 was published, Graham was diagnosed with lung cancer and wrote to Wynder stating that “there will be a tendency for some to say that my case tends to negate the idea that there is a causal connection between cigarette smoking and bronchiogenic carcinoma.”6 Graham's concern stemmed from the fact that he had quit smoking 5 years before his diagnosis and therefore, some individuals would argue that his smoking for a lifetime prior to that point was not the cause of his condition.
It was probably not until the publication of the US Surgeon General's 1964 report Smoking and Health that the link between smoking and lung cancer was broadly accepted.7 Therefore, acknowledgment that the Wynder and Graham1 study was published at a time when smoking was common and its negative health effects underappreciated justifies the categorization of the article by Wynder and Graham1 as classic in JAMA’s history and more broadly, in epidemiology's history.
The article by Wynder and Graham1 has many strong features that are exemplars of good study design and reporting. The authors thoroughly contextualized their findings, carefully cataloguing and crediting the work of others who came before them—most notably the studies that showed a temporal link between the increase in smoking prevalence and the subsequent increase in lung cancer death rates a few decades later. The authors also used great effort to limit the influence of recall bias and interviewer bias, repeating some of the surveys and blinding the interviewers to the study participants' case or control status. They also conducted several parallel data collections to ascertain the reproducibility of their results in different groups of patients.
The greatest strength of the study by Wynder and Graham1 may be the simplicity of the presentation. The authors used graphs, displaying unadjusted results by smoking exposure stratum, to illustrate that the likelihood of lung cancer diagnosis was higher in an individual with a heavy smoking history than in an individual without. The findings were strong and unequivocal, including a dose-response relation, in that the relative proportion of patients with lung cancer to control participants without lung cancer increased as the reported degree of smoking exposure increased.
Since publication of the article by Wynder and Graham,1 hundreds of studies have further analyzed the relation between smoking and lung cancer. Several studies suggested that the effects of smoking may differ between groups of patients. Black individuals may be more susceptible than white individuals to smoking's harmful effects.8 Longitudinal cohort studies have shown that women and men who have similar smoking histories appear to be at the same level of risk for lung cancer.9 Other investigations have attempted to characterize the interactions of age, sex, and smoking history to develop statistical models that could be used to predict an individual's risk of lung cancer in order to aid the design, implementation, and evaluation of lung cancer prevention and early detection studies.10
Wynder and Graham,1 along with many other researchers in the field, laid a foundation for tobacco control policy as a way of improving public health. The unveiling of cigarette smoking as a major source of disease and disability (probably the primary preventable source) serves as a clear example of how good science can influence public policy. The US Surgeon General's 1964 report on tobacco's numerous harms coincided with the high water mark for smoking in the United States (consumption peaked at roughly 4200 cigarettes, or 210 packs, consumed per adult per year).2 It has been estimated that cigarette use decreased immediately after the report's publication by almost 15% and a more durable 5% decline when measured over a period of several months.2 A steady stream of other public health interventions, ranging from the ban of broadcast advertising for cigarettes and the first Great American Smokeout in the 1970s, to the doubling of the federal cigarette tax in the 1980s, have been associated with a steady decline in smoking rates.
But there is still much room for individuals to conclude that science has not altered public policy enough. Cigarettes remain legal and widely available, although their negative health effects are well known and some hypothesize that cigarette use serves as a gateway to the abuse of other drugs and alcohol.11 There remains widespread frustration regarding the inconsistent and incomplete use of taxation to decrease consumption and initiation, despite decades of research demonstrating that price increases reduce cigarette consumption.12 These studies show that increases in the price of cigarettes stimulate quitting, reduce consumption by current smokers, and most important, sizably reduce the amount that young people smoke.
The Campaign for Tobacco-Free Kids13 reported that state cigarette tax rates varied widely in 2007 (the federal tax rate is $0.39 per pack sold) and many were low or effectively nonexistent. For instance in 2007, New York State imposed a tax of $2.75 per pack (New York City added another $1.50 per pack), but South Carolina imposed only a $0.07 per-pack tax rate.
Lack of consistency between neighboring states blunts the effects of higher taxes because of cross-border purchasing. For instance, in the District of Columbia and the state of Maryland, the tax rate is standardized at $2.00 per pack, but in the neighboring Commonwealth of Virginia, which shares a large border with Maryland and constitutes nearly half of the District of Columbia's border, the tax rate is only $0.30.
Public health advocates are also periodically disappointed that the proceeds from various successful lawsuits against tobacco companies have not been used more wisely. The Master Settlement Agreement is a case in point under which 7 major tobacco companies agreed in 1998 to pay more than $200 billion to 46 US states and 6 US territories to compensate for medical expenses incurred as a result of the use of their products.14 An analysis by Gross et al15 traced the states' use of these monies and found that in the sample year (2001), the average state received $28.35 in settlement funds per person, but applied only 6% of the funds toward tobacco control.15 This resulted in total average spending of $3.49 per person (partially funded by settlement dollars, partially funded from other sources), which was far less than the Centers for Disease Control and Prevention recommended that states direct towards tobacco control at the time.
Tobacco control policy will continue to be an important issue in the United States and abroad, with some unfinished business remaining on several fronts. For one, an existing World Health Organization treaty on tobacco control (the Framework Convention on Tobacco Control [http://www.who.int/fctc/en/index.html]) has been signed by 168 countries, but the United States has yet to ratify it. The Framework Convention on Tobacco Control articulates many of the basic principles shared by public health advocates and scientists regarding the harmful nature of tobacco and supply and demand reduction strategies. It includes price and tax measures to reduce demand for tobacco; nonprice measures, such as education and communication about tobacco's harms; and control of illicit sales of tobacco products and sales to minors.
In the United States, low levels of cigarette taxation and low levels of tobacco control program funding at the state level should be addressed. For instance, federal financing of health care to the states could be linked to those respective states' implementation of tobacco control strategies. One version of this strategy would be to link the federal medical assistance percentage to the extent of investment by the state in tobacco control programs, including the level of taxes levied. When states raise taxes in response to such an incentive program, they would garner more tax revenue from higher excise taxes and more federal support for their programs. As a third bonus, these states would incur lower direct medical costs due to the lower rates of smoking that would result. Such expansion of state programs would also ensure that current smokers and future potential smokers would have greater access to cessation aids and more financial incentives to avoid cigarettes. Almost 6 decades after Wynder and Graham published their classic study linking cigarette smoking to lung cancer, they most certainly would be happy to see these public health steps taken.
Corresponding Author: Peter B. Bach, MD, MAPP, Health Outcomes Research Group, Memorial Sloan-Kettering Cancer Center, 1275 York Ave, Box 44, New York, NY 10065.
Financial Disclosures: Dr Bach reports receiving consulting fees from Genentech, Wyeth, and Abraxis; and lecture fees from Genentech, Biogen-Idec, and Johnson and Johnson.
Bach PB. Smoking as a Factor in Causing Lung Cancer. JAMA. 2009;301(5):539-541. doi:10.1001/jama.2009.57