Prediction of chronic no-observed effect concentrations from acute toxicity data

Three methods are discussed to provide chronic estimates of lethal concentrations (LC) based on acute data. The methods use concentration time points rather than response to concentrations at one time as is commonly done with 96-h LC50s. The first method regresses LC values at each time on the reciprocal of time. The y intercept is interpreted as the LC under chronic exposure. The second method is the Multifactor Probit Analysis (MPA). Estimates of parameters in MPA models are determined by maximum likelihood. The equation is solved to forecast the LC at any time value. The third method predicts chronic toxicity based on the theory of ``Accelerated Life Testing`` (ALT). ALT was previously applied to mechanical failure under stress. The measurement taken is the time to failure (i.e. death). The first two methods statistically assume that observed response over a time series are independent. In many toxicity studies, this assumption may not be viable since the same experimental unit is observed repeatedly. This assumption of independence is not needed for the ALT method since the random variable is time to failure. All three methods are accurate in predicting chronic toxicity when compared to the results of actual chronic toxicity tests. Computermore » software was written for all three methods.« less