Software Glitch Slows PM NAAQS Review, August 2002
EPA has extended the schedule for the completion of the PM Criteria Document by one year, to October 2003, to allow for additional analysis of key studies.
The delay was necessitated by the announcement by Johns Hopkins University researchers on May 30, 2002 that misapplication of a common statistical software package led to changes in the quantitative results of the National Morbidity, Mortality, and Air Pollution Study (NMMAPS). The major qualitative conclusion — that there is strong evidence that particulate matter air pollution increases the risk of premature mortality — remains unchanged.
EPA has posted a memo about the issue, along with a letter from the Health Effects Institute, online.
EPA wanted to allow time for investigators of other studies to re-examine their results to determine if they were affected by
recommended changes to statisitical analysis techniques.
The scientific consensus is that the effect of the software error is likely to be greatest in the NMMAPS study because of its multifactoral design. The effects in other time-series studies are likely to be minimal. Case cross-over and cohort studies are not affected at all.
The schedule was discussed at an August 28 teleconference of the Clean Air Scientific Advisory Committee (CASAC), convened to consider EPA’s plan to secure timely reanalysis of key studies.
Prior to the meeting, EPA circulated some background materials for review, which are available online.
The EPA memo outlines:
1) Priorities for selecting studies for reanalysis. The criteria include: studies conducted in the U.S. and Canada; studies using measurements of PM2.5 and PM10-2.5, and some studies of PM10; multi-city studies; and single-city studies of special interest, such as those examining PM and co-pollutants, effects of PM components or use of factor analysis methods, distributed lag analyses, use of health endpoints other than mortality or hospital admissions, and assessment of harvesting and threshold issues;
2) A list of 29 priority studies for reanalysis based on the above criteria; and
3) A plan for facilitating the preparation and peer review of the reanalyses. EPA’s position is that it is the responsibility of the original investigators to perform the reanalyses, and the Agency will notify the authors of the affected studies. EPA will sponsor a technical workshop in late October to discuss results of re-analyses to data and to allow for mid-course corrections. A Health Effects Institute (HEI) panel will peer review will review the reanalyses.
4) A schedule. Reanlyses are to be completed by November or December, 2002, with an HEI “peer review” report due to EPA by late January 2003. EPA will issue a Fourth External Review Draft of the Criteria Document in March 2003. A CASAC meeting to review the draft will be held in June 2003, with a final PM Criteria Document completed in October 2003.
EPA indicated at the CASAC meeting that a draft of the Staff Paper should be available about 2 months after the publication of the 4th draft Criteria Document.
CASAC members suggested that EPA consult with some additional biostatisticians regarding the aspects of the statistical models recommended for re-analysis. They suggested that several additional studies be included on the priority list, and expressed concern that the proposed timeline may not be sufficient to complete the reanalyses.
Public health advocates have the responsibility to insure that industry does not use this delay as an opportunity to obfuscate the issues regarding the adverse health effects of particulate air pollution.
More information about the NMMAPS reanalysis, including a “FAQ” sheet, is available from Johns Hopkins University researchers online.
An American Lung Association fact sheet on the NMMAPS reanalysis is available online.
