Claire B. Ernhart Case Western Reserve University
Needleman's (1993) reply to our article in this journal ( Ernhart, Scarr, & Geneson, 1993) is a good example of the tactics he uses to deflect attention from questions of his scientific misconduct. Rather than address the many doubts about his scientific conduct, he attempted to focus readers' attention on (a) the motives and character of colleagues who question his research, (b) legitimate debates in the research literature on low-level lead effects on children, and (c) testimonials by colleagues who cannot know about misconduct in his research. He did not address our major thesis -- that procedures by which investigations of scientific misconduct are carried out require revision. Our experiences as whistleblowers were used to illustrate the need for change. Needleman's reply illustrates how he has dealt with the series of investigations of his scientific misconduct -- with sarcasm and innuendoes about the honesty and character of investigators, portrayed as conspirators with the lead industry (and newspaper columnists?) to bring him down. For more details of his obfuscation, see Scarr (1993), Ernhart (1993), and Ernhart and Scarr (in press).
It is remarkable that Herbert Needleman continues to attack us personally as whistleblowers in his case. The responsibility for evaluating the misconduct issue is not ours -- we have never had full access to the records. Our role was to report suspicions of his misconduct, described in the report that we filed with the National Institutes of Health-Office of Scientific Integrity (NIH-
Requests for reprints should be sent to Claire B. Ernhart, Department of Psychiatry, MetroHealth Medical Center, 2500 MetroHealth Drive, Cleveland, OH 44109-1998.
OSI) in May 1991. His colleagues at the University of Pittsburgh, not we, found him guilty of deliberate misrepresentations in his publications ( Needleman v. Healy, 1992; Cooley, Doreian, McCall, Reinmuth, & Rosenkranz, 1992). 1
From 1983 to the present, Dr. Needleman's continuing troubles with scientific integrity have resulted from increasingly thorough inquiries and investigations conducted serially by (a) the Expert Committee on Pediatric Neurobehavioral Effects of Low-Level Lead Exposure of the Environmental Protection Agency, (b) Drs. Ernhart and Scarr, (c) the NIH-OSI, (d) the Inquiry Panel at the University of Pittsburgh, (e) the Hearing Board at the University of Pittsburgh, and (f) the Department of Health and Human Services-Office of Research Integrity (DHHS-ORI, successor to NIH-OSI). At each stage, it was found that there were enough unresolved indications of irregularity to warrant continuance to the next step. The problems found now rest with the DHHS-ORI, for their judgment, and with the U.S. District Court for the Western District of Pennsylvania, which will have to resolve Needleman's suit against the University of Pittsburgh and the NIH.
At each step, the investigative groups could have found him innocent of misconduct and ended the series of investigations; they did not. Rather, the investigation process has continued serially because increasingly thorough investigations have documented more securely the actual conduct of his research. Despite his focus on our role in bringing suspicions about his research to the attention of the NIH-OSI, we are at least four steps removed from Dr. Needleman's current plight. The two University of Pittsburgh panels of dedicated and dispassionate scientists were the only ones with sufficient access to his data to make a limited but conclusive evaluation.
INVESTIGATION-UPDATE
At the request of the University of Pittsburgh, we kept the report ( Cooley et al., 1992) of the Needleman Hearing Board confidential while his appeals to his Dean and the Provost were reviewed. The Dean accepted the report, and Needleman subsequently withdrew the appeals to the Provost. The DHHSORI is now conducting its own investigation. If he is found to have committed scientific misconduct, Needleman will be offered a hearing. Because this hearing could be compromised by release of the report, we were asked to continue to maintain confidentiality.
Needleman v. Healy ( 1992 ), however, has made public an important find-
____________________
| 1 | The University of Pittsburgh released the Needleman Hearing Board Report ( Cooley et al., 1992 ) after the preparation of this article. This report, now publicly available, documents well both the substandard science and the deliberate misrepresentation involved in this study. |
ing -- He was found to have engaged in deliberate misrepresentation. How deliberate misrepresentation escapes the definition of misconduct is a matter of speculation ( Taylor, 1992). ORI recently reached a finding of scientific misconduct on the part of an investigator who was said to have "falsely reported" a critical fact and "intentionally mislead colleagues" (p. A1) in the Gallo case ( Hilts, 1992). If ORI concurs with the Pittsburgh verdict of deliberate misrepresentation by Needleman, it would be difficult to avoid a decision of misconduct in this case.
MORE EVIDENCE FOR THE ORI
ORI has access to more information than did the Pittsburgh Board. For example, a graph altered by Needleman may be included in the investigation. (The Hearing Board attributed the alterations to one of Needleman's associates. Needleman has not provided an explanation of these alterations and, even though he has used the altered graph in presentations, he is apparently allowing the attribution of responsibility to his colleague to stand.) The original graph, a plot of the cumulative frequency of Verbal IQ scores of low- and high-lead children, was deceptive in that no adjustment was made for confounding ( Needleman, Leviton, & Bellinger, 1982). The graph has since been enhanced ( Needleman, 1987) to bring the difference between medians to 6 points, similar to the difference between means that is central to Needleman's argument. Furthermore, additional high-lead, low-IQ points were added; this addition supports computation of high relative risk ratios for low IQ as a function of high lead level. This altered graph was used to support the recent redefinition of lead poisoning issued by the Centers for Disease Control ( CDC, 1991).
Lingering questions about reanalyses are highlighted by Needleman's reply. One reanalysis he has repeatedly attributed to the Environmental Protection Agency's (EPA's) statistician, Hugh Pitcher, evaporated at the Open Hearing, when Pitcher admitted he had never conducted a reanalysis of Needleman's data. Insofar as we can determine, all reanalyses, including those he provided to the EPA and published ( Needleman, Geiger, & Frank, 1985), were of various unspecified subsamples of less than the full sample of 270. Describing the most recent reanalysis, Schwartz (1993) stated that no subjects were excluded, but at Needleman's open hearing Schwartz said that he used the data of only 210 of the 270 children. Reports of this and the various other reanalyses provide no description of, or justification for, various subject inclusions and exclusions.
Needleman (1993) said, "each of these investigators was given a [italics added] set of raw data to use" (p. 93) for reanalysis -- a set, not the set of raw data that formed the basis of his disputed reports. From testimony at the Open
Hearing and from reports from the Inquiry Panel and Hearing Board, it is not clear Needleman gave the investigators the same raw data set that he used in his earlier analyses. In addition, reanalyses done by Needleman and by others were based on the same misrepresentations of subject selection and exclusion that tainted the original reports.
During the University of Pittsburgh's two investigations, Needleman did not reveal that his data were analyzed at least three times during data collection, that is, before the set of analyses we saw and before those provided to the Inquiry Panel. The Inquiry Panel was unaware of the existence of these earlier analyses. The Hearing Board discovered the second, but not the first and third, during its investigation. These three reports shed light on what was happening during data collection.
First, consider Needleman's statement that a decision to exclude certain classes of children was made a priori and that these criteria were applied through all analyses. This is not so; if these criteria had been set before the conduct of the study, it would have been apparent at the first analysis that considerable data to be excluded were being collected. The report of the first analysis ( Needleman, 1977) indicated that 157 children had been tested; 129 children were described, and the data of 93 were analyzed. This first report was quite specific regarding children not brought in for testing (bilingual home, moved, not interested, etc.) but did not report the exclusion of any already tested children. Of the 129 children described, 18 were of birth weight less than 88 oz (low birth weight).
In the second analysis ( Needleman, Gunnoe, Leviton, & Peresie, 1978), the data of 116 children were analyzed; children with low birth weight, history of lead intoxication, or head injury were excluded from analysis. Bilingual home was not mentioned.
In the third report ( Needleman, 1990), the data of 130 children were analyzed. Fifty-eight additional children were excluded on the basis of diagnosis as lead poisoned, bilingual home, or history of central nervous system injury. Bilingual home was now an exclusion criterion for already tested children, low birth weight was not. In this third report, the children were said to be excluded prior to scoring but their scores were entered into the computer. This report also noted that 47 children who had been tested and excluded for discrepancies in dentine lead level would be reconsidered for inclusion in the study if later dentine samples were concordant with earlier values.
Needleman (1993, p. 99) argued that a computer code used to exclude cases was based on preestablished criteria and not human judgment. The early analyses, with changing criteria for the exclusion of cases, indicate that he was well aware of the variables, and possibly the individual cases, that might affect findings. The exclusion of the "lead poisoned" cases (M IQ = 100, SD = 16) is deceptive. One of these children had an IQ of 134, yet Needleman has frequently stated that no high-lead child in the study had an IQ greater than
125. The reports of the Inquiry Panel and Hearing Board ( Cooley, 1992) documented continuing changes in the dentine lead criteria for inclusion in the study. The reports of early analyses indicate that the criteria for the exclusion of already tested children also changed during data collection.
The three analyses conducted during data collection were remarkable in other ways. In the third analysis (N = 130), the difference between the groups in reaction time was 0.0; in the published version (N = 158), the difference reached statistical significance. From the first to the second analysis the sample gained 21 low-lead, but only 2 high-lead children. From the second to the final sample, the increment was 17 low- and 25 high-lead cases. Testing of high-lead children late in the data collection period contributed to significant differences among groups in age, although Needleman categorically denied this bias in testing procedures at the Open Hearing.
More important, the selection of cases for testing may have otherwise been biased. Teacher ratings were available; the proportion of high-lead children rated hyperactive, and impulsive was significantly greater among the final 58 cases than the proportions of high-lead children so rated in the pool from which the sample was drawn.
Reports of the second and third early analyses indicated control for child's age; it was only in the analyses preceding publication that age was deleted. The effect of lead became statistically significant after this deletion. The ORI is aware of these new revelations.
LEAD INDUSTRY CONSPIRACY?
We state unequivocally that we are not employees or representatives of the lead industry or otherwise under any obligation to industry. Needleman's accusations amount to libel.
Needleman stated that within a year of the time that Ernhart ( Ernhart, Landa, & Schell, 1981) wrote "If there are in fact, behavioral and intellectual sequelae of low levels of lead burden independent of other aspects of parental and social influences on development, these effects are minimal," (p. 918) 2 she became a grantee of the International Lead Zinc Research Organization (ILZRO). This is false. The article was in press in 1980. Ernhart's ILZRO grant began in 1983. In the intervening time her lead-effects research was supported by the March of Dimes and the Perinatal Clinical Research Center (at Cleveland Metropolitan General Hospital), which, in turn, was supported by a U.S. Public Health Service grant.
____________________
| 2 | The italicized segment, important to the conclusion drawn, was omitted by Needleman. The inference drawn was determined by careful review of the results. This inference is not inconsistent with later research in which significant effects were found sporadically. When such effects are obtained, after control of confounders, they are small. |
Needleman writes as if there is something inherently suspect about receipt of an ILZRO grant. Given that such grants have no strings, it is entirely proper and beneficial to society for industry to contribute to the research enterprise. The money is not pocketed; as with other grants, it is extended to the grantee's institution to pay the expenses of doing research. A number of prominent researchers have received ILZRO support. Needleman often refers with admiration to his mentor, Randolf Byers, although Dr. Byers received research support from the Lead Industries Association (by ers, 1959; by ers & Maloof, 1954; Buyers, Maloof, & Cushman, 1954).
We have acted as expert witnesses in lawsuits involving purported lead poisoning. Defendants (primarily property owners, not industry) have the right to expert opinion. The fees we have charged are about 10% of what Needleman has been known to ask to testify for plaintiffs in the same kinds of cases. Lawsuits claiming lead poisoning have increased remarkably since the CDC (1991) redefined lead poisoning; Needleman played a major role on that committee.
Our attorney, David Geneson, is a member of a large, multi-city firm that represents more than 10,000 clients, so it is not surprising that Needleman finds some clients offensive. Actually, we contacted the firm because one of Scarr's daughters worked there.
LITIGATION TO PROTECT HIS DATA
Although Needleman (1993) may be technically correct in saying that he has not sued us before, it is not true that he "offered Scarr and Ernhart the opportunity to review my printouts in my lab" (p. 97). In fact, Needleman filed an affidavit in the U.S. District Court for the District of Utah to prevent our review of his data! The judge in the Superfund case ordered him to open his files to our scrutiny. This is a matter of public record.
Following the failed attempt to prevent us from seeing his data, the U.S. Department of Justice acted on his behalf (at taxpayer expense) to try to prevent us from divulging what we had seen in his laboratory. Presumably, the aim was to preserve his reputation as an expert witness in Superfund cases tried for the EPA. 3 We retained counsel to prepare and argue our case in order to preserve our right to freedom of speech. Fortunately, the judge ruled eloquently against Needleman, as quoted in our article ( Ernhart, Scarr, & Geneson , 1993).
Our reluctance to participate in the misconduct hearing was due to the suit to have us silenced, to Needleman's lawsuit against NIH and the University of
____________________
| 3 | Nothing about this matter was provided in materials obtained from the EPA under the Freedom of Information Act. |
Pittsburgh ( Needleman v. Healy, 1992), and, most important, to the fact that NIH and the University had accepted responsibility for the investigation.
MORE PARTICULARS
Needleman (1993, p. 98) complained that we were "shielded from answering certain questions" at his hearing. If the chairman was remiss, it was because he allowed Needleman's insulting and irrelevant questions to continue. The University's Research Integrity Policy states: "Reasonable rules of relevancy shall guide the chairperson in ruling on the admissibility of evidence."
Needleman contends that we apply a different standard to his work than to our own because we criticized him for dropping age as a confounder, although we did not include age in analyses in some of our own studies. This is specious. Age was considered but was not a confounder in our studies; it was in his. It was clear from his inclusion of child's age during early data analyses that he recognized the problem but, with child's age included, did not get the desired results.
At the hearing and in his response to us, Needleman has produced many personal testimonials by trusting colleagues and friends. With one exception, colleagues who worked with him on the study continue to be silent.
Needleman (1993) continued to contribute confusion to investigations of his work. One item in his reply to our article is new: The notion that the Inquiry Panel said that subjects who should have been included in the analyses were actually vacant case numbers and that he reported this to the Dean of the School of Medicine. Needleman's response (personal communication to G. M. Bernier , December 30, 1991) to the Dean says nothing about empty cases, and there is no mention of this notion in any other documents, confidential or not, available to us. Why would anyone enter nonexistent cases into computer data files?
CONCLUSION
Deliberate misrepresentation continues. Needleman attacks the whistleblowers, the NIH, and his own University to deflect attention from the indications of misconduct in his research. Federal agencies base policies on his research and on his assertions regarding the purported effects of low-level lead exposure on children. The policies divert resources and attention from the real needs of children. Misconduct issues will ultimately be resolved by the ORI and the courts. Public policy issues may require reconsideration in light of further evidence.
REFERENCES
by ers, R. K. ( 1959 ). Lead poisoning: Review of literature and report on 45 cases. Pediatrics, 23, 585-603.
by ers, R. K., & Maloof, C. ( 1954 ). Edathamil calcium-disodium (versonate) in treatment of lead poisoning in children. American Journal of Diseases of Children, 87, 559-569.
by ers, R. K., Maloof, C., & Cushman, M. ( 1954 ). Urinary excretion of lead in children. American Journal of Diseases of Children, 87, 548-558.
Centers for Disease Control. ( 1991 ). Preventing lead poisoning in young children. Atlanta: U.S. Department of Health and Human Services.
Cooley, W. W., Doreian, P. D., McCall, R. B., Reinmuth, O. M., & Rosenkranz, H. S. ( 1992, May 20 ). Needleman Hearing Board Final Report, University of Pittsburgh.
Ernhart, C. B. ( 1993 ). Deliberate misrepresentations. Pediatrics, 91, 171-173.
Ernhart, C. B., Landa, B., & Schell, N. B. ( 1981 ). Subclinical levels of lead and developmental deficit -- A multivariate follow-up reassessment. Pediatrics, 67, 911-919.
Ernhart, C. B., & Scarr, S. (in press). Substandard science, deliberate misrepresentation, and scientific misconduct. PSR Quarterly.
Ernhart, C. B., Scarr, S., & Geneson, D. F. ( 1993 ). On being a whistleblower: The Needleman case. Ethics & Behavior, 3, 73-93.
Hilts, P. J. ( 1992, December 31 ). Federal inquiry finds misconduct by a discoverer of the AIDS virus. New York Times, pp. A1, A11.
Needleman, H. L. ( 1977 ). Developmental effects of low level lead exposure. Unpublished manuscript (Grant Application No. HD 08945-04 to the Department of Health, Education, and Welfare).
Needleman, H. L. ( 1987 ). Low level lead exposure and children's intelligence: A quantitative and critical review of modern studies. Proceedings of the International Conference on Heavy Metals in the Environment (Vol. 1, pp. 1-8). Edinburgh: CEP Consultants.
Needleman, H. L. ( 1990 ). Deposition, exhib. 8, U.S. v. Sharon Steel, et al. Civ. Act. 86-C-924J (Utah, 1990 ).
Needleman v. Healy No. 92-0749, W. D. Pa. Filed March 25, 1992 .
Needleman, H. L. ( 1993 ) "Reply to Ernhart, Scarr, and Geneson". Ethics & Behavior, 3, 95-101.
Needleman, H. L., Geiger, S. K., & Frank, R. ( 1985 ). Lead and IQ scores: A reanalysis. Science, 227, 701-704.
Needleman, H. L., Gunnoe, C., Leviton, A., & Peresie, H. ( 1978 ). Neuropsychological dysfunction in children with "silent" lead exposure [Abstract]. Pediatric Research, 12, 374.
Needleman, H. L., Leviton, A., & Bellinger, D. ( 1982 ). Lead associated intellectual deficit. New England Journal of Medicine, 306, 367.
Scarr, S. ( 1993 ). A whistleblower's perspective on the Needleman case. Pediatrics, 91, 173-174.
Schwartz, J. ( 1993 ). Beyond LOEL's, p values, and vote counting: Methods for looking at the shapes and strengths of association. Neurotoxicology, 14, 237-248.
Taylor, R. ( 1992 ). Pitt's fuzzy verdict in the Needleman case. Journal of NIH Research, 4, 44.
