Dr. Carol Weisskopf, Analytical Chemist, WSU
Imagine that your child wants to go to a fairly strenuous and demanding summer camp. Before granting permission, you decide a medical examination would be a good idea. The child goes to the doctor, who orders a series of tests. The lab that is to perform the tests makes up numbers for the test results instead of doing the work. The doctor gets the bogus results, doesn't look at them, but states your child is fit for camp. The kid goes to camp and everyone is happy, right? Except that you've paid for a service you didn't receive, and you believe your child is healthy based on falsified data. Most of us would not knowingly tolerate such a cavalier approach to health decisions, and would not be pleased with either the lab or the doctor. One may think medical exams for summer camp are silly, but that doesn't imply approval of corruption in laboratory testing and medical review.
In this analogy, you can think of yourself as EPA, the doctor as a product registrant, your child as a pesticide (instead of a pest), and summer camp as an agrichemical dealership. The lab remains a lab, and what is described above is euphemistically referred to as dry labbing or pencil chemistry. Making up test results is generally considered bad form for grownups, and outside of an educational setting, is usually illegal as well.
Prior to the mid-seventies, EPA accepted laboratory data and reports pretty much at face value without misgivings about the quality of the data. This was a far cry from today's circumstances, when data quality is now documented in excruciating detail. EPA's Good Laboratory Practice (GLP) standards currently regulate facility operation, project conduct, and data documentation for nearly all pesticide studies submitted in support of product registration. GLPs have also spawned a new industry (GLP quality assurance professionals), give lots of people something to complain about, and seem to be widely detested. Implementation of GLP standards resulted in substantial modifications to operations in a wide variety of toxicology testing facilities, chemical analytical laboratories, and agricultural field stations. Just how did we get into this mess? It turns out the road to GLPs was paved with bad intentions.
The journey towards implementation of GLP standards got its first big impetus during facility reviews by the US Food and Drug Administration conducted in 1974 - 1975. The facilities conducted toxicology testing; the results were submitted to FDA in manufacturers' data packages. In 1975, Searle Laboratories became particularly noteworthy. A review of their facility demonstrated that test results were questionable due to sloppy work, untrained personnel, and poor data collection and analysis. Some test results were omitted from reports, and in many cases neither Searle nor the companies paying for the testing adequately reviewed data or reports.
Since initial review of testing facilities exposed significant defects, there was general agreement that guidance in proper performance standards was needed. A document called Good Laboratory Practice was developed and distributed. Industry compliance with these standards was voluntary; for most labs they would provide a reminder of what the agency expected from competent scientists. FDA also initiated a more formal inspection and audit program for study examination.
The next round of inspections, in 1975 - 1978, exposed two companies that made the Searle defects look trivial. Biometric Testing Incorporated and Industrial Bio-Test provided sufficient reason to promote Good Laboratory Practice from standard to law. Both companies routinely falsified test procedures and data, and provided fraudulent reports of test results. The Bio-Test animal facility was particularly damning: inadequate environmental control; dead animals unaccounted for; cages so insecure that animals could not only get out, but reenter cages for a different study group; and wild animals loose in the facility. Laboratory animals don't generally get visitation rights; poor animal tracking made study data useless.
Were these shortcomings sufficient reason to inflict a GLP mandate on us all? Industrial Bio-Test alone performed more than a third of all US toxicology studies. Documentation was frequently insufficient to distinguish between acceptable and unacceptable studies. Thousands of studies were invalidated, and several individuals were convicted on criminal charges. Along with pharmaceuticals, flawed testing encompassed more than 300 pesticides, bringing EPA into the picture. Faced with evidence of malfeasance of that magnitude, an enforceable regulation was a predictable agency response.
Final GLP rulings for toxicological research were published by FDA in 1978, and by EPA under TSCA and FIFRA in 1983. Those of us performing non-toxicological pesticide research continued on our casual way, but we didn't dodge the bullet for long. In 1989 EPA extended the coverage to virtually all agrichemical work under FIFRA, including field studies. In 1991 EPA issued a supplement to the FIFRA Enforcement Response Policy, describing responses for violations of the GLP standards. The described civil and criminal penalties made for sobering reading, and stimulated increased attention to the standards.
Another industry spawned by the GLPs was thus production of pocket-sized booklets containing the full text of the 1989 final ruling under FIFRA: Federal Register Vol. 54, No. 158; Part IV; Environmental Protection Agency; 40 CFR Part 160; Good Laboratory Practice Standards. Lab or field research directors seem to have dozens of copies; I think they're welded to the quality assurance folks. The standards are relatively brief, only 12 pages in a normal book. Unfortunately, a single sentence in the regulation can result in an extraordinary amount of work.
Given the nature of the fraud prompting the regulation, it is unsurprising that one focus is on record keeping. Work is now so thoroughly reported that it can be reconstructed, and deficiencies can be recognized. Compliance does not indicate good science. Well-documented poor science can conform to the standards (we can be idiots but not liars). It is undoubtedly still possible to cheat. Aside from the deterrence factor inherent in commission of a federal offense, under the regulations I think it would be both faster and easier to actually do the test than to do the paperwork required for plausible fraud. The documentation requirements, along with the corollary requirement for document retention, augmented another industry: document archiving.
Because many contract laboratories performing toxicological testing perform studies on chemicals destined for FDA and others for submission to EPA, they have been under some type of GLP regulatory standard since FDA's implementation of GLP in 1978. For non-tox studies, those of us in the agrichemical arena are still relative newcomers. Maybe we'll get used to it after we've been at it 20 years. One benefit to the GLPs is the existence of a quality assurance program in compliant laboratories, which tends to improve data quality. But, as with good science, quality assurance programs are not the exclusive property of GLP regulated labs.
I can disagree with the need for some required studies. I can believe that portions of the GLP standards are silly and that the regulations are overdue for revision. I can also resent usurpation of the title good laboratory practice. Is everything except 40 CFR Part 160 bad laboratory practice? But I can't say that the standards were an overreaction to finding out that nearly a quarter of the existing data was garbage. It's just that, after much dealing with GLPs, I want to go to summer camp.
Return to Table of Contents for the May 1998 issue