John Cleese’s 1970’s “Fawlty Towers” of BBC fame provided a satirical view of an inept hotel manager in his dealings with potential guests, “who is tortured by ‘that annoying section of the general public who insist on staying at hotels.’” Today, the torture is to that “annoying section of the general public” who insist on scientific integrity in the reports issued by the science community. Today’s “Fawlty Towers” are the “Faulty Ivory Towers of Science.”

A 2011 article in Scientific American titled “An Epidemic of False Claims” presents evidence of this general lack of scientific integrity in the biomedical field. More recently, the resurgence of measles illustrates the dangers to the public when medical research is taken to heart by the trusting public, and that research is later found to be “a fraud.” The prestigious medical journal The Lancet published a 1998 article by Andrew Wakefield claiming a link between the triple vaccine combination (measles-mumps-rubella) and inflammatory bowel disease and autism. An article by Brian Deer details the tortuous history of these claims leading to the eventual retraction of the article by the Lancet in 2010, and Dr. Wakefield’s removal from the U.K.’s doctor registry. With the seed of doubt of vaccine planted in the minds of anxious parents, many have since chosen to refuse vaccination for their children. The recent resurgence of measles in unvaccinated children demonstrates the public health risks resulting from incomplete vaccination of the childhood population. The controversy now continues into the realm of sound-bite political campaigning and pronouncements… an unwelcome development presaging the further loss of trust in both medical and political pundits.

Science has failed the public by not acknowledging adequately that research findings are always provisional awaiting some future refutation or affirmation, and that life can never be completely risk free.

Children and grandchildren have become the political tools in yet another area of scientific controversy… clean air. The U.S. EPA and activist environmental lobbying organizations have elevated dubious climate change concerns into emotional cries for the clean air well-being of one’s children and grandchildren. How clean is clean enough? Is it to be pure air; is it to be healthy air; safe air?  The exact level of air purity appears to be an ever-receding goal just over the next funding-grant horizon. Asthma rates have increased even as the air has become cleaner… never mind. EPA-acceptable, airborne particulate matter (PM) concentrations are being set ever smaller because computer programs have found observational correlations between air particulate levels and some parameter of health or disease. Never mind that the composition of the particulate matter in one part of the country may be quite different from that in a location thousands of miles away. Never mind that the pathophysiology for such a presumed linkage is lacking. Proposed EPA rules simply focus on particulate size concentration, such as PM2.5 (microns). Carbon dioxide has been targeted as harmful to humans; an EPA oversight is that we all normally exhale carbon dioxide at the four to five per cent level, as the EPA frets over parts-per-million.

A computer program can tease many “correlations” out of a sea of data. Some graphs show correlations blatantly meaningless correlations. One example shows that  per capita consumption of cheese (U.S.) correlates with the number of people who died by becoming tangled in their bed sheets. These two unrelated parameters appear follow each other in synchrony, yet they have no rational relationship.  Observational relationships require some testable hypothesis in order to tie the two variables together in a meaningful manner. Computer programs can be instructed to look for specific correlations amongst chosen parameters.  This differs from constructing a logically coherent hypothesis and then using a computer to provide supporting data points.

Comedian Lenny Bruce antedates the modern computer era, but one of his comedy routines illustrates the use of the computer to blindly follow instructions… with unintended consequences. A shopkeeper finds a genie’s magic lamp, and gets his three wishes granted. Left in charge of the shop, the unsupervised genie obligingly fulfills the wish of a hapless customer who asks “make me a malted.” The genie blindly obeys the command, and the customer is made into a malted. Modern computer data dredging programs obediently follow their master’s coded commands, and find what they are told to find.

Air quality issues come with prejudicial opinions as to health impacts. If the air looks dirty, smoggy brown, or is smelly, it is assumed that it has detrimental health effects on humans. Data pools are mined by computers designed to find the assumed correlations, and they find them...just barely. The strength of such correlations are quantified by such statistical measures as relative risk (RR), and a RR of 1 and a fraction is taken as valid proof of cause-and-effect by those researchers looking for confirmation of their favored assumption. Robust RRs are taken to be 2 or greater.

When the actual medical records of hospitalized patients are compared to prevailing ambient air-quality measurements there is no confirmation of presumptive ill-health effects. In 2008, University of California researcher James Engstrom’s analyses contradicted the accepted finding that diesel emissions were responsible for a big health toll in California. His reward was the threat by UCLA to terminate his employment, even though is research findings were not refuted.

Bio-statistician Stephen Milloy did a study of hospital admission records in central California and reported that: “Average ground-level ozone (O3) and fine particulate matter (PM2.5) measurements were not correlated with 19,327 patient admissions for asthma at the University of California-Davis Medical Center (UCDMC) during 2010-2012.” Examination of actual hospital records trumps blind data, and this study refutes EPA claims of increased hospital admissions related to air quality.

Similar commentary and elucidation of the statistical pitfalls of blind data mining is explained in a joint article by physician J.D. Dunn and S. Milloy. EPA’s reliance on computer mining in support of preconceived positions on air quality contrasts with traditional concepts of testable theory and rational biological associations to support such theories.

However, the EPA seems satisfied with Lenny Bruce’s genie’s task of “make me an association” willingly provided by the faulty ivory towers.

Charles G. Battig, M.D. , Piedmont Chapter president, VA-Scientists and Engineers for Energy and Environment (VA-SEEE). His website is www.climateis.com