Table 1 Characteristics and Comparisons of the Statistical Methods Used for Analysis of Biomarker Data with Limits of Detection
From: Toward improved statistical methods for analyzing Cotinine-Biomarker health association data
Method | n used (% total) | n below LOD | Transforms censored data to categorical data | Replaces censored values with imputed value | Inherently able to deal with multiple LOD? |
|---|---|---|---|---|---|
Complete Case | 5,865 (62) | 0 | No* | No* | No* |
Single Imputation | 9,488 (100) | 3,623 | No | Yes | No |
Multiple Imputation | 9,488 (100) | 3,623 | No | Yes | Yes |
Logistic regression | 9,488 (100) | 3,623 | Yes | No | Yes |
"Reverse" Kaplan-Meier | 9,488 (100) | 3,623 | No | No | Yes |