In the March 2000 SAP for the implementation of probabilistic risk assessments, one question posed to the panel concerned dermal toxicity estimation. The SAP was concerned about the potential approach of applying the ratio of oral-to-dermal toxicity from mammals to birds. A limited set of data exists for evaluating the relationship between dermal toxicity and oral toxicity in birds (Table H-1). Definitive oral and dermal LD₅₀'s are available for 42 individual studies. These studies were conducted across a variety of species and several classes of chemicals.

Regression analysis was used to predict the dermal LD₅₀ based on the oral LD₅₀. In addition, pesticide chemical properties were included into the regression model in an attempt to improve the model fit, similar to the approach taken by Mineau (2002). Prior to regression analysis, the dermal LD₅₀ data and the oral LD₅₀ data to were transformed to the log-base10 scale to better meet assumptions of normality and homogeneous variance.

In the log transformed scale, the correlation coefficient between the dermal and oral LD₅₀ values was 0.55 (Figure H-1). The correlation coefficients between dermal LD50 and the evaluated chemical properties were much lower: -0.11, -0.03, and -0.03 for molecular weight (MW), density, and molecular volume (MV), respectively.

Figure H-1. Plot of dermal LD₅₀ values vs. oral LD₅₀ values in log-base 10 scale. The correlation coefficient was 0.55.

A summary of the evaluated regression models is provided in Table H-2. This analysis indicates that the addition of these chemical properties into the regression model do not significantly improve its predictive ability. The adjusted R-square (a modified R-square with a correction for the number of parameters included in the model) ranged between 0.2575 and 0.2857, indicating little difference in the predictive ability among the fitted models. The best fitting model to predict dermal LD₅₀ was one based solely on oral LD₅₀:

log Dermal LD₅₀ = 0.84 + 0.62 * log Oral LD₅₀

standard errors: (0.14)(0.15)

The p-value for the slope was 0.0002 and the R-square was 0.30. The chemical properties that were included did not provide significant improvement in the predictive ability of the model.

Footnotes:

1. Hudson, R.H., M.A. Haegele, and R.K. Tucker. 1979. Acute oral and percutaneous toxicity of pesticides to mallards: Correlations with mammalian toxicity data. Toxicology and Applied Pharmacology 47:451-460.

2. Schafer, E.W., R.B Brunton, N.F. Lockyer, and J.W. DeGrazio. 1973. Comparative toxicity of seventeen pesticides to the quelea, house sparrow, and redwing blackbird. Toxicol. Appl. Pharmacol. 26:154-157.

3. Schafer, E.W. 1984. MRID 00146286

4. Schaefer, E.W. 1972. The acute oral toxicity of 369 pesticidal, pharmaceutical, and other chemicals to wild birds. Toxicol. Appl. Pharmacol. 21:315-330.

^b Data value was censored (50% mortality not obtained at highest dose) and was not used in the statistical analysis.

Literature Cited

Mineau, Pierre. 2002 Estimating the probability of bird mortality from pesticide sprays on the basis of the field study record. Environmental Toxicology and Chemistry 21:1497-1506.

Top of Page