Tests for strict monotonic trend in bio-medical dose-response relationships (respective concentration-response or exposure-response relationships) -- a biostatistical perspective

Abstract

Evidence of a global trend in dose-response dependencies is commonly used in bio-medicine and epidemiology, especially because this represents a causality criterion. However, conventional trend tests indicate a significant trend even when dependence is in the opposite direction for low doses when the high dose alone has a superior effect. Here we present a trend test for a strictly monotonic increasing (or decreasing) trend, evaluate selected sample data for it, and provide corresponding R code using CRAN packages.

11 Figures, 5 Tables