Exposing Defective Research, But Denied Credit for the Effort: Case 1

 

I have catalogued in this blog numerous instances of deficient research that has been published by respected medical journals.  In some cases, journal editors have published our concerns and authors’ corrections (here). 

It is extremely difficult to get critical letters published.  Editors don’t accept letters that simply complain about debatable issues.  Critics must raise concerns that are irrefutable and likely to affect the study’s results and/or conclusions.  Journal editors are ultimately responsible for failing to detect erroneous research, as they choose the articles and the peer reviewers.  When objecting to an article, the critic carries the burden of proof, often in the face of staunch, even coordinated opposition by all the other parties.

In this and subsequent posts I will describe cases in which editors accepted our criticism as legitimate, but defied medical publishing ethics by failing to publicly acknowledge our concerns, pretending that the errors were discovered by the authors.

________________________________________________________

Case 1. Mortality Study Among American Smokeless Tobacco Users by staff at the National Cancer Institute.

On May 17, 2019, Maki Inoue-Choi and colleagues at the National Cancer Institute and the University of Minnesota published a study, “Contemporary Associations of Exclusive Cigarette, Cigar, Pipe, and Smokeless Tobacco Use With Overall and Cause-Specific Mortality in the United States,” in the Journal of the National Cancer Institute Cancer Spectrum (here).  It asserted that smokeless tobacco (ST) use was associated with increases in “overall mortality (HR = 1.41, 95% CI = 1.20 to 1.66)” and “higher risk of mortality from heart disease and smoking-related cancer, with strong associations observed for cancers of the oral cavity and bladder.”

Working with the same data, my colleague Nantaporn Plurphanswat and I found that some of the Inoue-Choi results made no sense, with numbers in their tables not adding up, and tobacco use definitions being unclear.  Dr. Plurphanswat emailed Inoue-Choi with our concerns on July 31 (here).  A month passed with no response, so I emailed the senior author of the study on August 28 (here), and again received no response. 

Finally, on December 3, I emailed the editor of JNCI Cancer Spectrum with a summary of our five-month fruitless effort to contact the authors (here).  She responded the same day: “We may suggest you put this together as a letter to the editor that we will ask the authors to respond to.”

A week later, on December 10, an editorial assistant at the journal wrote: “Thank you for bringing this concern to our attention. We are proceeding with a correction to this article. We suggest no other action be taken until we publish the corrigendum.”

In February 2020, the journal published Inoue-Choi’s correction (here), which gives the false impression that they discovered their error.  Dr. Plurphanswat and I were not acknowledged for our weeks of work identifying those mistakes and raising our concerns with the authors.  More importantly, our other questions about the article went unanswered.  Furthermore, the correction itself was bizarre, as it listed the changes line by line in text that ran two pages longer than the original article.  A retraction and republication would have been more appropriate.

When editors allow authors to make mistakes and corrections this egregious, the scientific value of the finished product is seriously compromised. 

Authors and editors may try to evade a full accounting of the errors.  Further evidence of ethical lapses in publishing will be provided in Case 2 next week,

 

Influenza/Pneumonia Experience Offers Lessons for Covid-19

In only a few months, U.S. covid-19-related deaths surged past 100,000.

To put that in perspective one can examine the nation’s annual mortality rate for the similar diseases, influenza and pneumonia.  Together, these diseases are coded in the 10^th revision of the International Classification of Diseases as J09-J18, and, according to the CDC, the category is the 8^th most common cause of death in the U.S.  I downloaded the most recent ten years of data (2007 to 2016) from the CDC mortality website.  On average, influenza/pneumonia was responsible for 53,800 American deaths per year.  The exact number varied, from just over 50,000 in 2010 and 2012, to 57,000 in 2015. 

Covid-19 has been widely reported to pose an elevated risk for older adults.  In this regard it mimics the distribution of influenza/pneumonia.  In the chart above, note that 46% of influenza/pneumonia deaths were in the age group 85+ years, 26% were in 75-84-year-olds, and 13% in 65-74-year-olds.  In other words, a full 85% of deaths from influenza/pneumonia are in the retirement-age population.  These individuals frequently have underlying health conditions that make them more susceptible, or immune systems that are less robust than those of younger people.

However, no age group has complete immunity from influenza/pneumonia.  There are deaths at all ages, even though the numbers get very low.  The second chart below provides a closer look at ages from birth to 44 years.  Note that about as many children under 5 die from influenza/pneumonia as the entire group from 5 to 24 years. 

 

It will be a couple years before the covid-19 deaths and their distribution are understood in detail.  On March 24, the CDC assigned a new code for these deaths, assuring more accurate tracking.

Covid-19 is an entirely new infection, and the dynamics will not mirror influenza/pneumonia, for which many in our population are at least partly vaccinated.  However, policymakers should recognize that the most susceptible population is 65+ years old.  A question that should be debated: Is it possible to quarantine and protect the elderly while not shutting down the economy, the consequences of which are devastating, most severely for people with marginal incomes and resources?  Intelligent policymaking will require robust data and comprehensive analysis.

Finally, A Direct Comparison of Smoking and Smokeless Tobacco Use

I have documented how American health authorities refuse to directly compare the health effects of smoking and smokeless tobacco (ST) use.  The results of such an exercise would require them to acknowledge the products’ vast risk differentials.  For years, the American Cancer Society has possessed data that would allow this comparison (here, here, and here), but they refused to run the analysis or provide me with the data (here). I recently explained how FDA officials hid the comparison in a New England Journal of Medicine article (here).

I have spent much of the past 25 years trying to correct this information deficit.  Lacking access to the necessary data, the only comparison I could make was indirect (here), which was less than ideal.

Now, at last, the data are in full view.  Altria scientists in April published the first-ever follow-up mortality study of cigarette smokers and ST users, using national surveys and the National Death Index, all of which are produced by the U.S. Government and publicly available.  The first author of the impressive study, published in Harm Reduction Journal, is Michael T. Fisher. 

The figure at left illustrates the results for all causes of death, all cancers and heart diseases; smokeless tobacco is referenced as SLT.  In each section, hazard ratios – the likelihood of dying compared with never tobacco users – are illustrated for smokers by the first set of black dots/squares in the red circles; former smokers are in the next set; and ST users are in the third set, circled in blue.

Smokers are at more than twice the risk of dying from all causes than never tobacco users.  Former smokers’ odds are about 30% to 50% higher than those of never tobacco users (HR = 1.3 – 1.5).  Current ST users who never smoked died at the same rate as never tobacco users.

Compared with never users, smokers had even higher odds for dying from cancer, from 2.9 to about 4.2.  Former smokers also had higher odds, varying from 1.6 to 2.4.  Once again, ST users died at the same rate as never tobacco users.

Smoking isn’t as big a risk factor for diseases of the heart; other factors, like obesity, diet, physical fitness and diabetes, are also important.  Smokers in this study had odds ranging from 1.2 to 2.2, and not all of these were significant.  ST users had no excess risk.

In summary, this analysis of government data confirms that ST use is vastly safer than smoking.  The FDA and CDC not only had this data, but used it in other mortality studies of smokers and cigar users.  By not publishing the results on ST users, federal officials maintained the illusion that ST “is not a safe alternative to cigarettes.”  It is ironic that cigarette industry researchers produced this pivotal analysis.  Stay tuned to this blog for more results.