|
Correction
to
Needleman et al.,
|
| |||||||||||||||||||||||||||||
In my study, we used lead levels in teeth as the measure of lead exposure and attempted to collect multiple specimens from each subject. I reported that if three teeth from a given subject were analyzed, the lead levels in all three had to be in agreement for the subject to be included in the analysis. That was incorrect. That was the rule I had established at the beginning of the study, but I subsequently found that we were excluding too many subjects. I then established the more liberal rule that the values in two of three specimens had to agree. That was the rule we followed, but in writing the paper I mistakenly reported the original rule.
I also reported that the upper boundary of the initial dentine lead level for the low-lead group was 6 ppm of lead. This was an error. The correct figure was 8.7 ppm.
It is pertinent to note that these two statements were errors in reporting; they had no effect on the data analyses performed or the conclusions drawn. The classification of the lead level was performed without knowledge of the outcome, and the separation of the two groups according to the lead level was unequivocal. A high-lead classification required a mean value >20 ppm; a low-lead classification required a mean value <10 ppm. This was stated in the original report.
This article became the subject of charges of scientific misconduct brought against me by two defense witnesses for a lead smelter, in a lawsuit in which I was a government expert. An inquiry was begun at the University of Pittsburgh.
The University of Pittsburgh inquiry panel, which was assembled to determine whether there were adequate grounds to pursue an investigation, reanalyzed my data and found the same effect that I had reported. The panel claimed, however, that of the 36 eligible cases it had sampled and evaluated, 23, or 64 percent, were improperly excluded on the basis of dentine lead values. A careful review of these 23 cases in the original laboratory books showed that 18 did not meet the inclusion criterion for dentine lead and were properly excluded. Of these 18, 3 cases had no data on lead values at all; the relevant rows in the laboratory data books were empty. The five other cases were not in our telephone-contact file, indicating that we never spoke to the parents after the teeth had been analyzed.
The Office of Research Integrity also reanalyzed my data and found the same effects, but they charged that I had misrepresented the distribution of the children's IQ scores in a graph accompanying an article published in the Journal in 19822. The original graph had been hand-drawn by a medical artist. This is the graph in the top panel of Figure 1. In 1994 I entered all the original IQ data into my statistics software program, and the computer redrew the figure, which is shown in the bottom panel of Figure 1. I maintain that these two graphs display the same information.
|
Herbert L. Needleman, M.D.
University of Pittsburgh Medical Center
Pittsburgh, PA 15213
References
| |||||||||||||||||||||||||||||
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2008 Massachusetts Medical Society. All rights reserved. |
Previous
