LD50 power analysis

[Shiranai]

Hello all,

I come from a biology background, and I wanted to ask a question. We have a value of LD50 (lethal dose), which is the dose at which 50% of the animals die. We use this to test out chemicals.

The question that I have been given is to detect a drop of 100 fold in LD50 when a harmful chemical is given versus a nonharmful version. I am really not certain how to approach this question, but it appears that I should use some form of a student's t test for it. I would really appreciate any help because I am lost for answers. I suppose that I can just use some software afterwards, but really I have never taken any biostatistics courses before. Thanks!

 

[IronMan]

I think we may need more information to help you, as I don't completely understand your question.

"... to detect a drop of 100 fold in LD50 when a harmful chemical is given versus a nonharmful version. "

Are you talking about 1/100 of the dose? a 0.5% die off instead of 50% dieoff?

 

[Shiranai]

Sorry--that was worded funny. In other words, 50% mortality occurs at 1/100th of the dose (ie. 0.5 mg versus 50 mg) when the strong version is given (ie. harmful chemical). The mortality rate is "constant." The variances of the two treatments will likely be different as well if that factors into the decision.

I should have made it clearer if we have (in bacterial terms) superbug X and bug X. I need 100 times more bug X to kill 50% of the population.

These are assumed to be normal distributions by the way