A fraudulent study to advance political and research payola agendas deserves exposure and punishment.
The letter sent to the Office of Research Integrity is below. A PDF copy is here.
September 5, 2017
Dr. Kathy Partin
Office of Research Integrity
U.S. Department of Health and Human Services
Office of Research Integrity
1101 Wootton Parkway, Suite 750
Rockville, Maryland 20852
Re; Request for Investigation of Research Misconduct
Dear Dr. Partin,
I am requesting that the Office of Research Integrity (ORI) commence an investigation regarding research misconduct committed by the authors of the following study (“NEJM study”):
Air Pollution and Mortality in the Medicare Population. Qian Di, M.S., Yan Wang, M.S., Antonella Zanobetti, Ph.D., Yun Wang, Ph.D., Petros Koutrakis, Ph.D., Christine Choirat, Ph.D., Francesca Dominici, Ph.D., and Joel D. Schwartz, Ph.D. N Engl J Med 2017; 376:2513-2522 June 29, 2017DOI: 10.1056/NEJMoa1702747.
A copy of the study is attached. The reasons for the request are set out below.
I. ORI has jurisdiction in this matter as the NEJM study was funded by multiple grants from the Department of Health and Human Services.
The NEJM study was funded by the National Institutes of Health (Grant Nos. R01 ES024332-01A1, ES-000002, ES024012, R01ES026217) and the National Cancer Institute (Grant No. R35CA197449).
II. Misrepresenting research so it is not accurately represented in the research record is misconduct.
As the National Institutes of Health and the National Cancer Institute are parts of the Department of Health and Human Services, this matter is governed by the standards established in 42 CFR Part 93 — Public Health Service Policies On Research Misconduct. Thereunder, “research misconduct” means:
… fabrication, falsification, or plagiarism in proposing, performing, or reviewing research, or in reporting research results.
(b) Falsification is manipulating research materials, equipment, or processes, or changing or omitting data or results such that the research is not accurately represented in the research record…
(d) Research misconduct does not include honest error or differences of opinion. [Emphasis added]
The case is made below that the omissions in the case of the NEJM study not only misrepresent the research record but also are not the product of mere honest error or differences of opinion.
III.Facts: The NEJM study reports in no uncertain terms that PM2.5 causes premature mortality.
The NEJM study concludes in main part:
This study… showed that long-term exposures to PM2.5… were associated with an increased risk of death, even at levels below the current [regulatory standard]…
The overall association between air pollution and [premature mortality] has been well-documented since the publication of the landmark Harvard Six Cities Study in 1993.
The absolute certainty of these statements, made without qualification, inspired an editorial (attached) by the New England Journal of Medicine entitled, “Air Pollution Still Kills.” The editorial concludes with the sentence: “Do we really want to breathe air that kills us?”
Although the NEJM study authors carefully, if not cynically, used the term “associated with” rather than “causes,” there can be no doubt as to their intent to convey a false certainty that PM2.5 causes death.
IV. The researchers have committed misconduct by knowingly misrepresenting the research record.
A. No mention made of contradictory research.
The NEJM study authors failed to mention the existence of the contradictory findings of numerous other PM2.5-mortality epidemiologic studies despite knowledge by the authors/editors of their existence. Just some examples of recent significant contradictory findings include the following (Citation/Excerpt from Abstract/Comment):
- Young S et al. Air Quality and Acute Deaths in California. Regul Toxicol Pharmacol. https://doi.org/10.1016/j.yrtph.2017.06.003. (In press, online June 13, 2017). “Neither PM2.5 nor ozone added appreciably to the prediction of daily deaths. These results call into question the widespread belief that association between air quality and acute deaths is causal/near-universal.” Although this study became available at Regulatory Toxicology and Pharmacology in June 2017, it was first made available on Cornell University’s arXiv.orgweb site on February 10, 2015 (https://arxiv.org/abs/1502.03062). The study was also presented at a poster session at the 2016 annual meeting of the Health Effects Institute (HEI), one of the funders of the NEJM study.
- Enstrom J. Fine Particulate Matter and Total Mortality in Cancer Prevention Study Cohort Reanalysis. Dose-Response. http://journals.sagepub.com/doi/10.1177/1559325817693345. “No significant relationship between PM2.5 and total mortality in the CPS II cohort was found when the best available PM2.5 data were used.” Not only was this study published three months ahead of the NEJM study The editor-in-chief of the New England Journal of Medicine, Jeffrey M. Drazen, personally rejected the study for publication in the NEJM on June 28, 2016.
- Greven S et al. An Approach to the Estimation of Chronic Air Pollution Effects Using Spatio-Temporal Information. J. American Statistical Association. http://amstat.tandfonline.com/doi/abs/10.1198/jasa.2011.ap09392 (Published January 12, 2012).“[W]e are not able to demonstrate any change in life expectancy for a reduction in PM2.5.” One of the co-authors of this study, Francesca Dominici, is also a co-author on the NEJM study.
There are many other studies in the published literature that dispute the purported link between PM2.5 and premature morality. But the above-cited studies, in particular, were well known to those involved with the NEJM study. NEJM study funder HEI, NEJM study author Dominici and the NEJM study editor-in-chief Drazen all knew of these contradictory findings, yet there is still no mention or allusion to these or other studies in the NEJM study. This can only have occurred by design. The omissions cannot be viewed as inadvertent or honest error.
The NEJM study authors also omitted other key information that would have more accurately placed their results in the context of the research record.
B. The NEJM study authors omitted mentioning the limitations of epidemiology, including that there is no biological plausibility for the notion that PM2.5 kills.
Like all epidemiologic studies, the NEJM study is purely statistical in nature and cannot by itself establish a causal relationship between PM2.5 and premature death. As the U.S. Environmental Protection Agency (EPA), which is responsible for regulating PM2.5 in outdoor air, acknowledged to a federal court in litigation involving PM2.5:
[E]pidemiological studies do not generally provide direct evidence of causation; instead they indicate the existence or absence of a statistical relationship. Large population studies cannot assess the biological mechanisms that could explain how inhaling [PM2.5] can cause illness or death in susceptible individuals.
To assess the “biological mechanisms” that could explain how inhaling PM2.5 could cause death, animal toxicology or human clinical research is necessary. But none of the extant PM2.5 animal toxicology, human medical research or human clinical research studies supports the hypothesis that PM2.5 kills. In short, there is absolutely no physical evidence that supports the claim that PM2.5 kills.
In addition to the absence of biological, medical, or other physical evidence supporting the notion that PM2.5 in outdoor air kills, there is a host of real-world evidence ranging from the tobacco epidemiology to the epidemiology of workers with high exposure to PM2.5 (e.g., coal miners and diesel workers) to other high, real-world PM2.5 exposures (e.g., prior lethal air pollution incidents, ongoing high PM2.5 exposures in China and India, and forest fires) that plainly contradict the PM2.5-kills hypothesis.
The absence of physical evidence that PM2.5 kills has been admitted by the EPA in its explanation for conducting human experiments involving PM2.5. In explaining to a federal court why EPA researchers wanted to expose elderly human subjects to exceedingly high doses of PM2.5, EPA stated:
[Controlled human experiments] help to determine whether the mathematical associations between ambient (outdoor) levels of air pollutants and health effects seen in large-scale epidemiologic studies are biologically plausible (or not).
But none of the hundreds of EPA human study subjects exposed to PM2.5 has ever been harmed in the slightest by PM2.5.
In short, if PM2.5 kills anyone as the NEJM study authors claim to have demonstrated, no physical evidence of this phenomenon has ever been produced by anyone at anytime. The NEJM study authors failed to acknowledge this reality and its consequences for their dubious statistical results (discussed below).
C. The NEJM study authors misrepresented the interpretation of their statistical analysis.
The NEJM study relies on a statistical precision that simply doesn’t exist in real-world epidemiology because of unavoidable uncertainty surrounding the data. The NEJM study is a great example of the “garbage-in, garbage-out” phenomenon.
While the NEJM study purports to causally associate PM2.5 with premature mortality based on a hazard ratio on the order of 1.08, every epidemiologist knows that hazard ratios below the level of 2.0 are unreliable.
This is has been a long-held view maintained by bodies such as the National Academy of Sciences and National Cancer Institute, which stated in a media release on October 26, 1994:
In epidemiologic research, relative risks of less than 2 are considered small and usually difficult to interpret. Such increases may be due to chance, statistical bias or effects of confounding factors that are sometimes not evident.
In his highly-valued 1965 essay in the Proceedings of the Royal Society of Medicine, entitled “The Environment and Disease: Association or Causation,” Sir Austin Bradford Hill described the criteria for evaluating epidemiologic studies and discounted hazard ratios below 2.0:
First upon my list I would put the strength of the association. To take a very old example, by comparing the occupations of patients with scrotal cancer with the occupations of patients presenting with other diseases, Percival Pott could reach a correct conclusion because of the enormous increase of scrotal cancer in the chimney sweeps. ‘Even as late as the second decade of the twentieth century’, writes Richard Doll (1964), ‘the mortality of chimney sweeps from scrotal cancer was some 200 times that of workers who were not specially exposed to tar or mineral oils and in the eighteenth century the relative difference is likely to have been much greater.’
To take a more modern and more general example upon which I have now reflected for over fifteen years, prospective inquiries into smoking have shown that the death rate from cancer of the lung in cigarette smokers is nine to ten times the rate in non-smokers and the rate in heavy cigarette smokers is twenty to thirty times as great. On the other hand the death rate from coronary thrombosis in smokers is no more than twice, possibly less, the death rate in non-smokers. Though there is good evidence to support causation it is surely much easier in this case to think of some feature of life that may go hand-in-hand with smoking – features that might conceivably be the real underlying cause or, at the least, an important contributor, whether it be lack of exercise, nature of diet or other factors. But to explain the pronounced excess of cancer of the lung in any other environmental terms requires some feature of life so intimately linked with cigarette smoking and with the amount of smoking that such a feature should be easily detectable. If we cannot detect it or reasonably infer a specific one, then in such circumstances I think we are reasonably entitled to reject the vague contention of the armchair critic ‘you can’t prove it, there may be such a feature’.
The reason hazard rations below 2.0 are unreliable is because much epidemiologic data are incomplete, guesstimated and/or otherwise of dubious validity. The unreliable data problem is writ large in the NEJM study:
- No information on cause of death. The NEJM study data lacks information on the cause of death for any individual in the Medicare population ⎯ so deaths not possibly caused by PM2.5 (e.g., those resulting from accidents, homicide/suicide, cancer, etc.) are included in the study population.
- Guesstimated exposure data. The NEJM study relies entirely on guesstimated exposure data extrapolated from relatively few air monitor measurements. These guesstimated data have no relationship to actual PM2.5 exposures among the study subjects which are affected in the short-term and long-term by occupational, residential and lifestyle PM2.5 exposures that are not measured by outdoor air monitors. Smokers in particular inhale thousands of time more PM2.5 from tobacco than they inhale from outdoor air. In studies like the NEJM study-touted Harvard Six City Study, about 50% of the study population are either current or former smokers. In these cases, PM2.5 exposures from outdoor air pale in comparison and are insignificant to PM2.5 exposures from smoking. Attribution of death to PM2.5 in outdoor air is an exercise in statistical absurdity.
- Confounding risk factors ignored. The NEJM study fails to consider confounding factors such as smoking, socioeconomic status and any of the other myriad potential competing risk factors for death. In essence, the NEJM study assumes all “excess” deaths are PM2.5-related.
A particularly egregious example of the NEJM study authors’ failure to consider confounding risk factors occurred a mere two weeks after the NEJM study was published. On July 13, 2017, the NEJM published another study from Harvard School of Public Health researchers reporting that poor diet was associated with premature mortality. Despite the near simultaneity of this study with the NEJM study, the authors of the NEJM study did not consider diet as a potential confounding factor for mortality. Both studies involve Harvard School of Public Health researchers studying the same health endpoint (premature mortality) and published by the same journal (New England Journal of Medicine), but neither study considers other study’s exposure of concern as a confounding factor in its own results. Are we really to believe this failure was inadvertent?
Also, the NEJM study authors repeatedly present their hazard ratio estimates as “risk” estimates. It is “Epidemiology 101” that, despite terminology like “relative risk,” hazard ratios are not estimates of risk. Hazard ratios are merely measures of the statistical correlation between exposure and health endpoints in specific study populations. This “strength of association” measurement may then be used along with all the (Bradford Hill) criteria in determining whether actual cause-and-effect can be identified. But hazard ratio estimates have nothing to do with risk per se. Communicating hazard ratios as risk is deceptive.
D. NEJM study authors misrepresent the Harvard Six Cities Study.
As cited above, the NEJM study authors base the credibility of their results on the allegedly “landmark Harvard Six Cities Study of 1993.” In addition to the fact that the Harvard Six Cities Study is yet another dubious piece of statistics-only work, the co-authors of that study have hidden their data from outside/independent scrutiny for about 23 years.
The EPA’s Clean Air Act Scientific Advisory Committee, Congress and qualified researchers have made multiple requests for the raw data underlying the Harvard Six Cities Study. All requests have been refused by the study authors.
One of the Harvard Six Cities Study researchers refusing to make this data available for independent replication is NEJM study co-author Joel Schwartz.
Between its secret data and dubious epidemiologic analysis, the only things “landmark” about the Harvard Six Cities Study is the study authors’ ability to hide their data for more than 20 years and their sheer arrogance in then offering it up as validation of the NEJM study claims.
It is also worth mentioning that NEJM study co-author Antonella Zanobetti is also a data-hider. She has also refused to provide PM2.5-related study data to qualified researchers for purposes of study replication.
E. EPA compelled NEJM study author forced to recant negative PM2.5 study results.
EPA once compelled NEJM study author Francesca Dominici to recant negative PM2.5 study findings. Unhappy with the EPA-funded 2011 Greven et al study contradicting EPA’s PM2.5-kills claims on which Dominici was a co-author, EPA pressured Dominici to explain them away. Dominici complied in writing (letter attached and highlighted in relevant part) by nonsensically stating that while her study showed PM2.5 did not kill on a local level, her study showed that PM2.5 killed on a broader national level. This is patently absurd. If PM2.5 causes death as hypothesized, then it causes death everywhere.
F. Peer review or “pal” review?
There is no doubt that the NEJM study authors will raise peer review as a defense to these charges. This is an entirely bogus defense. I have attached a copy of a recent Wall Street Journal op-ed explaining how the PM2.5 “peer” review process is more like “pal” review.
As an example, Harvard University’s Doug Dockery sits on the EPA scientific advisory committee responsible for “peer” reviewing the EPA-funded Harvard Six City study, for which he was also the lead author. Reviewing your own work is not “peer” review. It is likely that the “peer” reviewers of the NEJM study are either:
- Fellow PM2.5 cronies of the study authors; or
- Lack familiarity with the PM2.5 epidemiology and controversy.
So there was no legitimate peer review of this study.
G. Political nature of the HSPH/NEJM study.
Given the current political situation ⎯ a new administration reportedly looking to cut EPA’s budget (including for university-conducted research into PM2.5) and cut EPA’s regulatory overreach — the political nature and timing of the HSPH/NEJM study and editorial cannot be overlooked.
The study result is not novel. The editorial drives home a wild political attack on President Trump, concluding with the irresponsible implication that President Trump’s administration is going to cause U.S. air to be polluted to lethal levels — i.e., “Do we really want to breath air that kills us?”
It is worth noting that while air pollution did kill people on several occasions during the 20th century, these deaths were NOT caused by particulate matter but by temperature inversions that trapped and concentrated emissions of caustic gases.
In an interview about the NEJM study, NEJM study author Francesca Dominici told the media that:
We are now providing bullet-proof evidence that we are breathing harmful air.
So the intent of the NEJM study authors is clear ⎯ to present their study as incontrovertible evidence that PM2.5 kills. They attempted to accomplish this by intentionally omitting from their study key information that entirely contradicts and deflates their claim. Theirs is a deliberate attempt to misrepresent the research record. This is a fraud on the government and taxpayers who have funded this “research.” These researchers should be appropriately sanctioned.
Finally, in the event that you disagree with any or all of these allegations, I request a detailed response explaining your specific points of disagreement.
Please let me know if you require further information.
Steven J. Milloy