Accommodating Measurements Below a Limit of Detection: A Novel Application of Cox Regression

• Published in: American journal of epidemiology Vol. 179; no. 8; pp. 1018 - 1024
• Main Authors: Dinse, Gregg E., Jusko, Todd A., Ho, Lindsey A., Annam, Kaushik, Graubard, Barry I., Hertz-Picciotto, Irva, Miller, Frederick W., Gillespie, Brenda W., Weinberg, Clarice R.
• Format: Journal Article
• Language: English
• Published: Cary, NC Oxford University Press 15.04.2014; Oxford Publishing Limited (England)
• Subjects: Biological and medical sciences; Biomarkers; Biomarkers - blood; Case-Control Studies; Clinical outcomes; Cohort Studies; Computer Simulation; Confounding Factors, Epidemiologic; Cross-Sectional Studies; Data Interpretation, Statistical; Developmental Disabilities - blood; Developmental Disabilities - chemically induced; Dioxins; Environmental Exposure - analysis; Environmental Pollutants - blood; Environmental Pollutants - toxicity; Epidemiologic Research Design; Epidemiology; Female; General aspects; Health Surveys; Human exposure; Humans; Infant; Limit of Detection; Linear Models; Logistic Models; Male; Mathematical models; Medical sciences; Metabolic diseases; Miscellaneous; Models, Theoretical; Odds Ratio; Polychlorinated Biphenyls - blood; Polychlorinated Biphenyls - toxicity; Polychlorinated Dibenzodioxins - blood; Practice of Epidemiology; Proportional Hazards Models; Prospective Studies; Psychomotor Disorders - blood; Psychomotor Disorders - chemically induced; Public health. Hygiene; Public health. Hygiene-occupational medicine; Regression analysis; Sex Factors; Nutrition; Nutrition survey; hazard identification; Regression; Identification; limit of detection; Epidemiology; 2,3,7,8-tetrachlorodibenzo-p-dioxin; proportional hazards; Feeding; 2,4,4'-trichlorobiphenyl; nondetects; Limit; Detection; National Health and Nutrition Examination Survey; Hazard; Public health; 2,4,4′-trichlorobiphenyl

In environmental epidemiology, measurements of exposure biomarkers often fall below the assay's limit of detection. Existing methods for handling this problem, including deletion, substitution, parametric regression, and multiple imputation, can perform poorly if the proportion of "nondete...