Binary regression for risks in excess of subject-specific thresholds

Biometrics. 1999 Dec;55(4):1247-51. doi: 10.1111/j.0006-341x.1999.01247.x.

Abstract

We describe models for binary valued data to be used to explain the incidence of disease given the level of a known risk factor. Every individual has an unobservable tolerance of the risk. Risk levels below the individual tolerance do not increase the disease incidence above the background, unexposed rate. We estimate parameters from both the tolerance distribution and the risk function for a large group of mice exposed to very low levels of a known carcinogen.

MeSH terms

  • 2-Acetylaminofluorene / administration & dosage
  • 2-Acetylaminofluorene / toxicity
  • Animals
  • Biometry*
  • Carcinogens / administration & dosage
  • Carcinogens / toxicity
  • Drug Tolerance
  • Female
  • Humans
  • Liver Neoplasms, Experimental / chemically induced
  • Mice
  • Models, Statistical
  • No-Observed-Adverse-Effect Level
  • Regression Analysis*
  • Risk*
  • Urinary Bladder Neoplasms / chemically induced

Substances

  • Carcinogens
  • 2-Acetylaminofluorene