Assays with lower detection limits: implications for epidemiological investigations

Area of Science:

• Environmental Health
• Biomarker Analysis
• Epidemiology

Background:

• Chronic low-level exposures pose challenges in epidemiological studies due to difficulties in detecting and quantifying biomarkers.
• Instrumentation limitations, including detection failures below a certain value and interference from noise, complicate accurate biomarker measurement.
• Current laboratory practices often use a limit of detection (LOD) based on instrument noise, categorizing values below this as 'not detected'.

Purpose of the Study:

• To discuss the application of detection thresholds in biomarker measurement within epidemiological investigations.
• To illustrate how conventional thresholding methods may inadequately serve epidemiological research.
• To advocate for a more nuanced approach to handling biomarker data near detection limits.

Main Methods:

• Review of conventional laboratory practices for determining the limit of detection (LOD).
• Analysis of how standard thresholding procedures impact the interpretation of biomarker data in epidemiological contexts.
• Discussion of the implications of treating data below the LOD as 'not detected'.

Main Results:

• Conventional LOD determination focuses on instrument noise and avoiding false positives, potentially creating a false dichotomy in data interpretation.
• Standard thresholding may fail epidemiological investigations by not fully utilizing available measurement information.
• Observed data below the threshold are often discarded or reported as 'not detected', losing valuable information.

Conclusions:

• Conventional approaches to biomarker detection limits can be insufficient for epidemiological studies of chronic low-level exposures.
• Investigators should request all observed biomarker data, including values below the standard threshold, from laboratories.
• A thorough understanding of the measurement process is crucial for appropriately treating and interpreting low-level biomarker data in epidemiology.