Problems with the Draize Test
Stephen R. Kaufman, M.D.
As an ophthalmologist, I strongly support use of alternatives to the
Draize test to determine oculary irritancy of cosmetic and household
products. The Draize test is scientifically unsound and inapplicable
to clinical situations. Reliance on this test is in fact dangerous, because
the animal data cannot be reliably extrapolated to man. Substances "proven" safe
in lab animals may in fact be dangerous to people.
As a Chief Resident, I have three years of experience at Bellevue Hospital (an Eye Trauma Center), where I have treated scores of toxic eye injuries in the emergency room. I have never used Draize data to assist the care of a patient. Furthermore, I know of no case in which another ophthalmologist found Draize data useful.
Ocular irritancy testing is often performed in order to label substances accurately as toxic or non-toxic. Here, the Draize test, due to its scientific inadequacies, fails miserably. The Draize test uses rabbits because they are inexpensive, have large eyes, and are easy to handle. However, the rabbit is an inappropriate and inaccurate model for human ocular damage. The following are some of the fundamental anatomical differences between the rabbit and human eyelid, tearing mechanism, and cornea:1-6
- The rabbit epithelial (surface) layer is 10 times
more permeable to hydrophilic solutes than the human
- Bowman's membrane (the next layer) is six times
thicker in man.
- The rabbit's threshold of pain in the eye is much
higher than that of humans, so irritating substances
are not washed away as readily.
- Rabbits have a less efficient tearing system than
- Unlike people, rabbits have a nictitating (winking)
membrane (third eyelid), which has an unclear effect
on elimination of foreign materials.
- Humans develop corneal epithelial vacuoles in response
to some toxic substances, but rabbits do not.
- The rabbit mean corneal thickness is .37 mm, while
that of man is .51 mm.
- Rabbits are more susceptible to damage (alkaline)
materials, because the pH of their aqueous humor
is .82 compared to .71-.73 for man.
- The cornea represents 25% of the rabbit eye surface area, but only 7% of the surface area in man.
Due to these differences, Draize data correlates poorly with actual human experience. Indeed, the limited available human data has demonstrated the inadequacy of the Draize test. Freeberg et al.7 reported 281 human ocular toxicity exposures to 14 household products, and they compared the findings to Draize test results. The human experiences differed from the Draize results by a factor of up to 250. The closest correlation differed by a factor of 18. Furthermore, the severity of rabbit eye response predicted poorly the degree of human ocular injury. The correlation coefficient between rabbit and human response was only 0.48 with p > .1. Thus, the Draize test predicts human eye toxicity poorly. Indeed, Griffith and Freeberg wrote:
The widely used Draize/FHSA rabbit eye irritation test has never been validated against any reported human database. As an in vivo surrogate for predicting human ocular response to irritants, it has been soundly criticized on both technical and humane grounds...8
The poor performance of the Draize test and the fundamental anatomical
differences between the human and rabbit eye highlight the shortcomings
of an in vivo test versus an in vitro alternative. While
the in vivo model will ways be compromised by the anatomical and
physiological differences between the experimental animal and man, in
vitro technology can improve as better tests are developed. Thus,
we should encourage the refinement of vitro technologies by eliminating
those in vivo tests which have been shown to be inferior or inadequate.
When no acceptable in vitro test exists, we could continue animal
tests while searching for better alternative tests.
Elimination of the Draize test in favor of alternatives would encourage greatly the development of alternatives in all areas of toxicity testing. The in vitro technology already allows a battery of tests which compare favorably with the Draize test. Shopsis et al. found a correlation coefficient of .84 for cytotoxicity assay in an external validation program.9 Cytotoxicity tests have great potential. As Maurice notes, "It is not unlikely that direct cytoxic action of agents on the cornea is almost entirely nonspecific and that cells of many different types would respond very similarly to those of the cornea if their exposure to the agent could be matched."10 Leighton et al. have developed a chick embryo model, which tests for the immune component of irritation.11 The EyetexTM system, using a chemical reagent of macro-molecular solutes, recorded the same irritancy classification as the Draize test for 61 of 67 tested chemicals, and the other six were within one Draize irritancy classification. There was an overall correlation of 96%.12
Given the inadequacy of the Draize test and the demonstrated reliability of alternative assays, it is unfortunate that the Food and Drug Administration has opposed legislation against the Draize test. Hertzfeld and Myers observed:
...if all testing were to shift to in vitro assays, then many firms now geared to animal testing would find their labs practically useless... The status quo is a strong motivator for those now profitable firms. Regulatory authorities tend to follow, not lead in accepting new technologies, and they are heavily influenced by the industrial concerns now in place.13
Progress in product safety testing will not come from over-reliance on outdated animal models. The Draize test was never and will never be a reliable irritancy assay. As modern technologies are developed, it can no longer be considered a "gold standard." I support the continued use of animal toxicity tests when there are no adequate alternatives. However, if we fail to eliminate the Draize test when compelling scientific evidence supports such a move, then enthusiasm for developments of all alternative technologies will wane.
1. Clelatt, KN (Ed): Textbook of Veterinary Ophthalmology. Lea & Febiger,
2. Prince JH, Diesem CD, Eglitis I, Ruskell GL: Anatomy and Histology of the Eye and Orbit in Domestic Animals. Charles C. Thomas, Springfield, 1960.
3. Saunders LZ, Rubin LF: Ophthalmic Pathology in Animals. S. Karger, New York, 1975.
4. Swanston DW: Eye irritancy testing. In: Balls M, Riddell RJ, Warden AN (Eds). Animals and Alternatives in Toxicity Testing. Academic Press, New York, 1983, pp. 337-367.
5. Buehler EV, Newmann EA: A comparison of eye irritation in monkeys and rabbits. Toxicol Appl Pharmacol 6:701-710:1964.
6. Sharpe R: The Draize test-motivations for change. Fd Chem Toxicol 23:139-143:1985.
7. Freeberg FE, Hooker DT, Griffith JF: Correlation of animal eye test data with human experience for household products: an update. J Toxicol-Cut & Ocular Toxicol 5:115-123:1986.
8. Griffith JF, Freeberg FE: Empirical and experimental bases for selecting the low volume eye irritation test as the validation standard for in vitro methods. In: Goldber AM (Ed): In Vitro Toxicology: Approaches to Validation. New York, Mary Ann Libert, 1987, pp. 303-311.
9. Shopsis C, Borenfreund E, Stark DM: Validation studies on a battery of potential in vitro alternatives to the Draize test. In: Goldberg AM (Ed): In Vitro Toxicology: Approaches to Validation. New York. Mary Ann Liebert, 1987, pp. 31-44.
10. Maurice D: Direct toxicity to the cornea: a nonspecific process? In: Goldberg AM (Ed): In Vivo Toxicology: Approaches to Validation. New York. Mary Ann Liebert 1987, pp. 91-93.
11. Leighton J, Nassauer J, Tchao R, Verdone J: Development of a procedure using the chick egg as an alternative to the Draize rabbit test. In: Goldberg AM (Ed): Product Safey Evaluation. New York. Mary Ann Liebert, 1983, pp. l65-177.
12. Gordon VC, Bergman HC: The EYETEX-MPA system. Presented at the Symposium, Progress in In Vitro Technology, Johns Hopkins University School of Hygiene and Public Health, Baltimore, Maryland, November 44, 1987.
13. Hertzfeld HR, Myers TD: The economic viability of in vitro testing techniques. In: Goldberg AM (Ed): In Vitro Toxicology. New York. Mary Ann Liebert, 1987, pp. 189-202.