Please quote ref no: SC/ST - L8/826
22 May 1998
Dr Kay H. Jones
Suite 18
2600 Fairview Ave. East
Seattle, Washington, 98102>br> USADear Dr Jones
I have now received a readable copy of Dr Davis' reply to your letter. I am passing this on to you. She answers all your points very fairly. Our decision is against publication in the journal, however.
Yours sincerely,
/s/
Stephanie Clark PhD
Senior Editor
JONES COMMENTS REGARDING POPE ET AL (1995) [by Devra Davis]
"Our paper, presented at that conference, showed that Pope's findings were suspect because available data for several cities were excluded form the analysis. We performed our own analysis with some additional and some common cities and found no statistical association between PM 2.5 and mortality."
The study by Pope and colleagues (1) used data from a prospective mortality study of approximately 1.2 million adults to identify deaths occurring between 1982 and 1989. The authors used air pollution data on fine particle concentrations that were available from 1979 to 1983 in 50 metropolitan areas to assess long-term pollution exposure. Using the Cox Proportional Hazards model, the relative risks of all-muse and cardiopulmonary mortality for the most polluted areas compared to the least polluted areas were 1. 17 (1.09-1.26) and 1. 3 1 (I. 17-1.46), respectively; these risk ratios were adjusted for age, sex, race, cigarette smoking, exposure to passive cigarette smoke, body-mass index, drinks per day of alcohol, education, and occupational exposure.
The analysis by Jones et al. in an unpublished monograph sidestepped the major analyses done by Pope using the prospective cohort data, instead focusing on a much more simple ecologic analysis depicted in supplemental figures in the paper. These plots used area wide unadjusted mortality for 1980. In the extension of these simple plots, Jones included the 'additional' cities by using air quality data that were collected between 1987 and 1993 . As stated in a response from Dr. Pope to EPA regarding this approach (2): 'This is a very strange suggestion. The additional AIRS pollution data that he suggests that we should have used were collected after the occurrence of most of the deaths, at different sites, and for only a few cities.' Further, Jones and coworkers also ignore the more robust ecological results for a far larger number of cities obtained by Pope et al. for a second indicator of fine particles, atmospheric sulfates.
The analysis by Dr. Jones and colleagues has not been published in a refereed journal, and details on how well the mortality information used compared even with the ecologic analysis by Pope is not readily available. What information is available in the form of the Citizens for a Sound Economy Foundation papers, fact sheets or press releases does not provide convincing evidence that the findings of Pope and colleagues are invalid. In this regard it is important to note the general agreement between Pope et al. (1995) with respect to mortality and indicators of particulate pollution in a second recent prospective cohort study (Dockery et al. 1993) as well as with earlier published ecologic analyses of the 1980 mortality data set by Ozkaynak and Thurston (1987).
"We also showed that EPA's mortality estimates were inflated because EPA did not note that Pope mistakenly used 'median' rather than 'mean' particulate data. EPA acknowledged our finding and reduced its annual mortality estimate by 25%, from 20,000 annual deaths to 15,000."
While a typographical error on one of the figures in the Pope study (1) led to overestimates in EPA's draft regulatory impact analysis, this in no way calls into question the findings reported by Pope and colleagues. As stated by Dr. Pope (2): 'Jones correctly notes a typographical error in figure heading 2 where the word mean should have read median. Please note, however, that the fine particulate air pollution index was accurately reported and used as median fine particulate air pollution in the text. Also median fine particulate air pollution was correctly used in the Cox Proportional HazArd analysis, the ecologic regressions, in creating the plot, and in interpreting the hypothesis tests conducted in the analysis.' More importantly for present purposes, EPA staff ' promptly corrected the analysis and the approach used in Davis et al. recognizes the correct specification of the air quality data. Hence the comment regarding a problem in an earlier draft EPA analysis is irrelevant.
"Moreover, EPA's estimate of 15,000 deaths includes 14,000 that are estimated to occur among people exposed to less than what EPA set as the standard which the EPA presumed to have as a reasonable margin of safety, i.e., a median of 15 ug/m3 or a mean of 18 ug/m3. When we assume that there is no discernable premature mortality,below the assumed threshold level, the number of premature deaths dropped to less than' about the Pope study i.e., no statistical association between PM 2.5 and mortality."
The assumption that there is no discernable premature mortality below 18 ug/m3 annual mean is not defensible. The absence of any obvious threshold in observed concentration-response relationships is a hallmark of the numerous recent epidemiological studies on the effects of particulate matter. Jones and colleagues are at least correct in discussing an 'assumed' as opposed to a 'demonstrated' threshold. Moreover, even in assuming a threshold level, Jones et al. are in error regarding the standard promulgated by EPA. The July 18, 1997 Federal Register clearly states that the standard is 15 ug/m3 annual mean not a median. 'Furthermore, Jones own estimates submitted in comments to EPA suggest that, in the U.S., a substantial fraction of the estimated numbers of fives prolonged would come concentrations exceeding 15 ug/m3 annual mean. Because the U.S.has generally cleaner air that found in major cities in much of the rest of the world, the existence of a possible threshold at 15 ug/m3 is unlikely to make a great difference in the results in the Davis et al. paper.
References:
(1) Pope CA III, Thun MJ, Namboodiri W Dockery DW, Evans JS, Speizer FE, Heath CW Jr. 1995. Particulate air pollution as a predictor of mortality in a prospective study of U.S. adults. Am J Respir Crit Care Med 151:669-674
(2) Pope CA III. 1997. Memorandum to John Bachmann, April 14, 1997. U.S. Environmental Protection Agency, Washington DC. Docket No. A-95-54, IV-E-61
(3) Post E. 1997. Memorandum from Ellen Post to John Bachmann and Colleagues, June 6, 1997. U.S. Environmental Protection Agency, Washington DC. Docket No. A-95-54, IV-A-8
(4) EPA. 1997. Regulatory impact analyses for the particulate matter and ozone national ambient air quality standards and regional haze rule. July 17, 1997, Innovative Strategies and Economics Group, Office of Air Quality Planning and Standards, U.S. Environmental Protection Agency, Research Triangle Park, NC.
Comments on this posting?
Click here to post a public comment on the Trash Talk Bulletin Board.
Click here to send a private comment to the Junkman.
Copyright © 1998 Steven J. Milloy. All rights reserved on original material. Material copyrighted by others is used either with permission or under a claim of "fair use." Site developed and hosted by WestLake Solutions, Inc.
Material presented on this home page constitutes opinion of Steven J. Milloy.