Optimal cutoff point

[Ryan Jones]

Hello,

I'm trying to figure out how to calculate a change in cut-off score for a screening test when you know the change in positive results but not their disease status. Currently Im using roctab to assess the sensitivity and specificity of the cut-off points but am unsure about how to go about calculating this.

Thank you

[Clyde Schechter]

It sounds like you are attempting to do the impossible. What you describe falls far short of what is needed to optimize a cutpoint. To do that you need to know:

1. sensitivity and specificity of the test operated at each cutpoint.

2. the overall prevalence of the condition that the test targets.

3. the losses associated with false positive and negative results.

From what you describe, you are missing the last two.