Confounding By Infection

"Omission is the most powerful form of lie." - George Orwell

The anti-smokers' so-called "mountain of evidence" about the supposed health risks of smoking has been created by systematically either ignoring the role of infection, including the suppression of research, and/or by using statistical methods that generate bogus risks by residual confounding.

Cigarette smoking as a potential cause of cervical cancer: has confounding been controlled? AN Phillips, GD Smith. Int J Epidemiol 1994 Feb;23(1):42-49.

This is a landmark paper, raising issues that the health establishment refuses to address -- except with their familiar evasive techniques of the curt dismissal, the stonewall, and the conspiracy of silence. The implications of this paper do not just threaten their claims about smoking and cervical cancer. Other infectious pathogens fit the model of Phillips & Smith as well, and not just sexually transmitted ones. All of those diseases in which there is a socioeconomic differential in exposure (which includes most of them), in which proxy measures are used as the measures of exposure to the etiological pathogen, have the same defect. And it is not just smoking, but many of the health establishment's sacred cows about "healthy lifestyles," such as diet and exercise, which are imperiled.

Say Phillips & Smith: "It is widely believed that some sexually transmitted pathogen plays a key role in the aetiology of cervical cancer. However, although certain human papillomaviruses are strongly implicated, the pathogen responsible and the mechanism of action remain to be finally characterized. The correlation between cigarette smoking and sexual activity that exists in most cultures therefore makes evaluation of the potential additional role of smoking difficult, due to confounding with the presence of the aetiological pathogen. Epidemiological studies of the association between smoking and cervical cancer have adjusted for the lifetime number of sexual partners as a proxy measure of the presence of the pathogen and, in most cases, the association has diminished but remained statistically significant. Since the use of a proxy will tend to result in underestimation of the effect of the aetiological pathogen on risk of cervical cancer, however, the adjustment is likely to be insufficient, thus resulting in an overestimation of the adjusted or 'independent' effect of smoking. In an attempt to address this concern we used a simulation approach to investigate whether a substantial 'independent' association between smoking and cervical cancer might be expected as a result of the use of a poor proxy for the aetiological pathogen, even if there is no true effect of smoking. Using realistic estimates of the association between the presence of the aetiological pathogen and both smoking and risk of cervical cancer, 'independent' relative risks for cigarette smoking of two and above were generated. It is therefore plausible that the observed 'independent' effect of cigarette smoking on cervical cancer is due to residual confounding."

Phillips & Smith - Int J Epidemiol 1994 abstract / PubMed

When their paper was written, an odds ratio of around ten was considered the best current estimate for the risk of human papillomavirus in cervical cancer. Since then, improvements in the technical methods of detecting HPV have led to its detection in virtually 100% of cervical cancers -- and the OR for HPV has soared to over 350. This high odds ratio has made the presence of confounding in all studies claiming to find a smoking risk a certainty. A high odds ratio has an even more powerful effect than a lower one.

Odds ratios of over 100 are common in infectious diseases. In many of them, no other cause of the disease can be credibly claimed to exist. so, in studies claiming to find a smoking risk, the proxy measure that is exploited to generate confounding is that of the amount of exposure to the etiological pathogen. They use indirect approximations such as income, number of family members, education, and employment to pretend that they have "adjusted" for any supposed "effects" of these variables (which are really just as fictitious as the alleged smoking risk; i.e., the lack of a high school diploma does not cause cancer!). But they have not adjusted the differential out, and are making false representations. The old cervical cancer studies (and some new ones) were "adjusted" in this manner, and they still failed to eliminate the bogus smoking risk.

And it is not just human papillomavirus which has been proved over time, through the improvement of methods of detection, to be a more important risk factor than originally thought. The same process has taken place with other pathogens, such as hepatitis B virus and liver cancer. But still, the health establishment continues to deceive the public with confounded studies, falsely blaming smoking for diseases it does not cause!

Bosch et al confirmed Phillips & Smith's model

From the International Journal of Epidemiology, 1994 Oct;23(5):1100-1101:

From FX BOSCH, S DE SAN JOSE AND N MUNOZ (letter):

Sir -- In the article by Phillips and Smith "Cigarette smoking as a cause of cervical cancer: Has confounding been controlled? it is argued that the increase in risk of cervical cancer associated with smoking observed in some studies could well be explained by an insufficient adjustment by some measures of exposure to the aetiological pathogen. Recent developments have shown that the major risk factor of human cervical cancer is human papillomavirus (HPV). We have completed five case control studies on cervical cancer where the prevalence of HPV DNA had been assessed using polymerase chain reaction (PCR). Our results show that the association between HPV and cervical cancer is very strong with odds ratios ranging from 15 to 100. Many of the 'traditional' risk factors strongly associated with cervical neoplasia in either pre-invasive or invasive forms, such as the number of sexual partners or age at first sexual intercourse were not associated with cervical cancer among women who were HPV DNA positive, while the association persisted among the HPV DNA negative. In our data the association between smoking status and cervical cancer the association was weak with odds ratios ranging from 1.4 to 2.0 after adjustment for major confounders including HPV status. However, when HPV positive women were examined, thus removing those with potential underdetected HPV, the odds ratios of cervical cancer in five data sets (3 on invasive cancer and 2 on CIN II lesions) were consistently and statistically not different from 1.

"Real data are thus in agreement with the model proposed by Phillips and Smith."

Bosch et al. (letter and reply) Int J Epidemiology 1994 / UCSF (pdf, 9 pp)

See Also:

The Percentage of Cancer Caused By Infection

The health establishment exploits confounding

The health establishment is fully aware of the possibility of confounding, and knowingly uses it to deceive the public about the health risks of active and passive smoking. In 1989, a study in the Journal of the American Medical Association proclaimed that "Cigarette smoking and exposure to passive smoke are risk factors for cervical cancer" (ML Slattery et al. JAMA 1989 Mar 17;261(11):1593-1598). It claimed that the adjusted risk estimate for active smoking was 3.42 and for passive smoking was 2.96. This study did not even mention the possible role of human papillomavirus, never mind attempt to detect it, and all the adjustments were via surrogate variables. The media were jubilant at having another defamation to sling. It was only in the editorial comment on the study (which the media ignored) that th issue of confounding by HPV was addressed (Smoking and cervical cancer: cause or coincidence? PM Layde. JAMA 1989 Mar17;261(11):1631-1633.

"The fundamental concern of the skeptics is what epidemiologists call confounding. In the case of cigarette smoking and cervical cancer, this refers to extraneous differences that exist between smokers and nonsmokers with respect to exposure to other risk factors for cervical cancer. Even the best efforts to control for reported sexual activity and other risk factors are limited in their ability to correct the problem of confounding."

"It is widely assumed that the most important cause of cervical cancer is an infectious agent that is transmitted sexually. Although herpes simplex virus type 2 was once thought to be a prime candidate for being the causative agent, human papillomavirus is now considered more likely on the basis of several recent studies. As a group, women who smoke appear to be more sexually active than women who do not smoke and, thus, may be more likely to be exposed to a variety of sexually transmitted diseases, including the causal agent for cervical neoplasia. Since this agent has not been unequivocally identified, however, it is not possible to control for differences in exposure to the causal agent between smokers and nonsmokers. The measures of sexual activity that have been used in epidemiologic studies are likely to be only moderately correlated with actual exposure to the specific causal agent, therefore, controlling for indexes of sexual activity will only imperfectly adjust for the true confounding influence of the causal agent, resulting in inflated 'adjusted' relative risks."

"The ultimate resolution of the causality of the association of passive and active cigarette smoking and cervical cancer may have to wait identification of the sexually transmitted infectious agent that appears to be the most important etiologic factor in cervical carcinogenesis. Current research on human papillomavirus appears very promising, but definitive clarification of the role of human papillomavirus will probably require time-consuming prospective studies. Development of an accurate and convenient diagnostic test for exposure to the true etiologic agent would permit tight control for its real confounding influence in epidemiologic and clinical studies and, hence, allow more careful elucidation of the contributions of factors with weaker associations, including active and passive smoking."

Layde continued: "Fortunately, definitive resolution of this issue is not urgent from the standpoint of public health or preventive medicine. It is clear that women should quit smoking cigarettes for many reasons other than a possible increased risk of cervical cancer." And, Layde openly hoped "that the issue of passive smoking might be of only historical interest before its role in cervical carcinogenesis is totally clarified."

It clearly doesn't matter to Layde if nonsmokers as well as smokers died unnecessarily from cervical cancer caused by HPV while waiting for the outcome of such leisurely investigations. Anti-smokers could (and did) simply pretend that they were done in by a whiff of secondhand smoke instead of a virus, and they used this lie to help manipulate the public to support smoking bans. Nor does it matter to Layde that those "many reasons other than a possible increased risk of cervical cancer" for the sake of which people are supposed to quit smoking include many that are just as fraudulent as the claim that smoking causes cervical cancer.

But as if to slap Layde and his ilk in the face for his smug callousness, within only a few short years it had been proven that human papillomaviruses are the cause of virtually all cervical cancer, and the mechanisms by which they did so were identified. Furthermore, it hadn't required "time-consuming [and expensive] prospective studies," thus foiling those villains' scheme to milk the taxpayers as well. It bears mentioning that the unadjusted supposed smoking risk in the Slattery study 10.1, which shows that confounding can produce more than just weak or moderate bogus risks. And as for those supposed passive smoking risks which are so grossly out of proportion to the relative eposures to passive versus active smoke, these only serve to provide evidence that active smoking itself plays no role, and the only relevant exposure is to HPV.

Another Deliberate Fraud

Although their pretense that smoking causes cervical cancer is untenable, the health establishment continues to attempt to find something to blame on smoking anyhow. And they continue to use proxy variables in multivariate analyses to purposely exploit residual confounding in order to do so, e.g., the recently-ballyhooed study by Moscicki et al. (Risks for incident human papillomavirus infection and low-grade squamous intraepithelial lesion development in young females. A-B Moscicki, N Hills, S Shiboski, K Powell, N Jay, E Hanson, S Miller, L Clayton, S Farhat, J Broering, T Darragh, J Palefsky. JAMA 2001 Jun 20;285(23):2995-3002). In this study, their pretense that controlling for number of sexual partners is appropriate is false. The number of sexual partners is not in itself a risk, it is the number of INFECTED partners that produces the risk. The pretended smoking risk of 1.67 (95% CI 1.12 to 2.48) is merely residual confounding from what Moscicki admits to be an "astronomically" high risk of HPV infection with each new sexual partner (10.10, 95% CI 3.24 to 31.50). The purported lack of association between sexual behavior and LSIL is likewise the fruit of confounding, which can result in false non-associations as well as false associations, exactly as described by Phillips & Smith.

Moscicki - JAMA 2001 abstract / PubMed

The "Federal Reference Manual on Scientific Evidence" Embraces Statistical Fraud

The 1994 Reference Manual on Federal Evidence of the Federal Justice Center, which is intended to educate the US Government's judicial branch employees in the wake of the Daubert decision, embraces the defective statistical technique of multivariate analysis. It is falsely claimed that "Multivariate analysis controls the confounding factor through mathematical modeling," which it does not. The paper by Phillips and Smith unequivocally proves that this technique generates bogus risk factors when the true, causal risk factor(s) have not been fully and accurately detected, and there is a noncausal association between smoking and this true risk factor.

From the "Reference Guide on Epidemiology," IV. General Causal Association Between Exposure and the Disease, A. Could a Confounding Factor Be Responsible for the Study Result? 2. What techniques, if any, were used to control confounding factors, page 160 (PDF page 40):

"2. What techniques, if any, were used to control confounding factors?

"To control for confounding factors during data analysis researchers can use one of two techniques: stratification or multivariate analysis.

"Stratification reduces or eliminates confounding by evaluating the effect of an exposure at different levels (strata) of exposure to the confounding variable. Statistical methods can then be applied to combine the different results of each stratum into an overall estimate of risk. For example, in McMahon's study of smoking and pancreatic cancer, if smoking had been a confounding factor, the researchers could have stratified the data by creating subgroups based on how many cigarettes each subject smoked a day (e.g., a nonsmoking group, a light smoking group, a medium smoking group, and heavy smoking group). By comparing the different rates of cervical cancer for people in each group who drink the same amount of coffee, the effect of smoking on pancreatic cancer is revealed. The effect of the confounding factor can then be removed from the study results.

"Multivariate analysis controls the confounding factor through mathematical modeling. Models are developed to describe the simultaneous effect of exposure and confounding factors on the increase in risk. This technique relies on building a series of mathematical models to predict who will get the disease. For instance, MacMahon might have begun a multivariate analysis with a simple model to determine how well the individual's daily intake of coffee predicts whether he or she will contract pancreatic cancer. In the next model, he could add the number of years the person had been a coffee drinker. If the second model better predicts who would contract cancer, MacMahon would continue to create more complex models (including variables such as age, gender, and ethnic group) until he found a model that best predicts who will contract cancer.

"If the association between exposure and disease remains after completing the assessment and adjustment for confounding factors, the researcher applies the guidelines described in section IV.B to determine whether an inference of causation is warranted.

Reference Guide on Epidemiology (First Edition, 1994) / US Federal Justice Center (pdf)

In the Second Edition, Gordis re-wrote parts of the chapter. The offending section is re-titled "3. What techniques can be used to control for confounding factors?" The example with MacMahon is eliminated, and a new paragraph is added: "Both of these methods allow for 'adjustment' of the effect of confounders. They both modify an observed association to take into account the effect of risk factors that are not the subject of the study and that may distort the association between the exposure being studied and the disease outcomes." The new version also falsely implies that multivariate analysis eliminates confounding.

Reference Guide on Epidemiology (Second Edition, 2000) / US Federal Justice Center (pdf)

This procedure of multivariate analysis is exactly what the health establishment robotically performs in studies of cervical cancer and other supposed "smoking-related disease." It generates fraudulent risks for non-causally associated risk factors. Note that stratification will not eliminate confounding when exposure to a true, causal, infectious risk factor has not been fully measured, either. Confounding occurs when the data from the different strata are combined. Bosch et al confirmed that confounding had occurred in real data by removing HPV negatives with potential underdetected HPV from the analysis. This is the proper way to stratify. In contrast, the health establishment would analyze the supposed HPV-negatives, and falsely claim the existence of a smoking risk among them, when these are really false negatives.

The medically qualified author of this "Reference Guide on Epidemiology" is Leon Gordis, M.D., Dr.P.H., Professor, Department of Epidemiology, Johns Hopkins School of Hygiene and Public Health, and Department of Pediatrics, Johns Hopkins School of Medicine, Baltimore MD. Leon Gordis is a crony of Jonathan Samet, the anti-smokers' star perjuror in the Minnesota tobacco trial, at the Center for Epidemiology and Policy at The Johns Hopkins School of Public Health. This is the edition of the Manual in effect at the time of the state tobacco trials.

The Center for Epidemiology and Policy, The Johns Hopkins School of Public Health
The Jonathan M. Samet Page

Leon Gordis' Trail of Lies

From 11-1-93 to 6-30-95, Leon Gordis received $376,501 from the anti-smoking Robert Wood Johnson Foundation's Clinical Scholars Program. "This is a program to train post-doctoral fellows from different medical/surgical specialty fields in effectiveness research, quality-of-life research and evaluation of community interventions."

Gordis's funding from the Robert Wood Johnson Foundation

Gordis is "lead faculty" of the Foundation for American Communication (FAC), which claims to "provide reporters with teaching from some of the nation's top academic experts" -- such as anti-smoker Stanton Glantz (!), Nov. 6-8, 1998.

Re: FACS Nov 6-8, 1998 / Junkscience.org

Leon Gordis was also one of the "Speakers on Scientific, Technical, and Specialized Knowledge Areas" at the ALI-ABA professional course, "New Directions in Expert Testimony. Scientific, Technical and Other Specialized Knowledge Evidence in Federal and State Courts," May 4-5, 2000, at the Ritz-Carlton in Boston.

"New Directions in Expert Testimony" / ALI-ABA

Leon Gordis was a consultant to Ernst L. Wynder's American Health Foundation journal, "Preventive Medicine," in 1984.

The American Health Foundation

"The Reference Manual on Scientific Evidence is the product of a cooperative effort by the Federal Judicial Center and the Carnegie Corporation of New York."

"The Center began its work on a manual to help federal judges deal with scientific evidence in 1990, shortly after the Federal Courts Study Committee recommended the preparation of such a manual." David A. Hamburg, president of the Carnegie Corporation of New York, William T. Golden and Joshua Lederberg, co-chairs of the Carnegie Commission, and David Z. Robinson, executive director of the Carnegie Commission, headed the effort. (Preface, p. vii.)

Preface, Reference Manual on Scientific Evidence / Federal Judicial Center (pdf)

Among the reviewers of the draft were C. Thomas Caskey, of the Washington Advisory Group; Gilbert S. Omenn; W. Kip Viscusi, a favorite public pretender of the tobacco industry; and Lee Loevinger of Hogan & Hartson, who had been an FCC Commissioner when it concocted its outrageous "Fairness Doctrine" to serve anti-smoker lawyer John Banzhaf. And, "The names of several reviewers have been omitted from the list at their request."

List of Reviewers, Reference Manual on Scientific Evidence / Federal Judicial Center (pdf)

The Reference Manual was just part of a larger, internationally coordinated effort by the Carnegie Group, which included a secret meeting with the scientific advisors and science ministers of all the developed countries.

The Framingham Heart Study Started This Fraud

From: Mapping the World of Epidemiology, Part 2. The Techniques of Tracking Down Disease. By Eugene Garfield. The Scientist 1988 Sep 5;11(36):290-295. (Garfield is a member of the Board of Directors of the Lasker Foundation's biggest lobbying group, Research!America.) "The publication of some of the first results of what is now commonly referred to as 'The Framingham Study' in 1971 represented another advance in the statistical methodology of epidemiology. Published by William B. Kannel and colleagues, Heart Disease Epidemiology Study, Framingham, Massachusetts, and the National Heart and Lung Institute, NIH, it has been cited in nearly 850 publications. The study, started in 1949 and still ongoing, is an investigation of the effects of a large number of variables on the risk of developing coronary heart disease, the number-one killer in the US. In his Citation Classic commentary, Kannel states that the 1971 study 'was one of the largest bodies of data showing the impact of cholesterols and lipoproteins on risk using prospective data.' The techniques of multifactorial analysis developed for this study have revolutionized the analysis of epidemiological data. We identified this study in an analysis of highly cited papers from the Annals of Internal Medicine."

The Framingham Heart Study is the apple of the Lasker Syndicate's eye. Mary Lasker and Florence Mahoney took over this study when they took over the National Heart and Lung Institute, and used it to propagandize for the pseudo-science of John Harvey Kellogg, at whose Institute in Battle Creek Mahoney had been a student. Thanks to them, research on the role of infection was suppressed for over 30 years. Also thanks to their immense power and control, this fraudulent methodology was shoved down the nation's throat without scientific dissent or Congressional oversight. And it has produced vast misallocation of resources and uncounted millions of needless deaths, as it serves the social engineering goals of its health fascist advocates.

Garfield / Garfield Library (pdf)
How the Public Was Brainwashed About Heart Disease

The "Sacred Lie" of Health Fascism -- McGinnis & Foege 1993

(Actual causes of death in the United States. JM McGinnis, WH Foege. JAMA 1993 Nov 10;270(18):2207-2212.) McGinnis & Foege's "Actual Causes of Death in the United States" is adored by health fascist bureaucrats. The supposed smoking deaths actually come from the CDC's SAMMEC computer program. The deaths blamed on toxic agents, primarily pollution (perhaps they should have called them "Miasma deaths"), and those blamed on diet and lack of exercise, are vastly inflated by blaming them for deaths caused by chronic infection. But, deaths from microbial causes (which are 4% of their total) parsimoniously include only those deaths due to acute infection, and are further reduced by not including HIV and "those otherwise estimated to be attributable to tobacco use, alcohol use, sexual behavior, and illicit use of drugs." For example, only 25% of deaths from primary liver cancer are attributed to the hepatitis B virus. And never mind that sex in the absence of pathogens causes death in virtually no one.

McGinnis & Foege - JAMA 1993 abstract / PubMed
McGinnis & Foege, JAMA 1993 / UCSF (pdf, 6 pp)
The Percentage of Cancer Caused By Infection

Class Conditions Amplify Exposure to Infection

Passengers in economy class are more likely to be infected by H1N1, and it is more easily transmitted in that class as well: "The risk of infection increased with the flight duration in both first class and the main cabin, but was consistently greater among economy-class travelers, investigators reported online in BMC Medicine. The increased risk in the economy cabin owed primarily to the 'more crowded conditions in economy,' Brian J. Coburn, PhD, of UCLA, said in a statement. 'Unfortunately, there is a very high probability -- 75% -- that if an infected person is on board, they will be in the economy cabin.' The findings extend patterns demonstrated for other viral infections, including measles, smallpox, tuberculosis, SARS, and seasonal influenza. Previous studies had shown that virus associated with each condition could be transmitted during commercial flights." (To Reduce H1N1 Risk, Fly First Class. By Charles Bankhead. MedPage Today, Jan. 7, 2010; re: Calculating the potential for within-flight transmission of influenza A (H1N1). BG Wagner, BJ Coburn, S Blower. BMC Med 2009 Dec 24;7(1):81.)

To Reduce H1N1 Risk, Fly First Class, 2010 / MedPage Today
Wagner - BMC Med 2009 abstract

<= HOME

cast 01-08-10