Table 2 Misclassification among USTUR Registrants.

From: Inference under outcome misclassification in health risk models using a simulation study with a validation dataset

Calculation method

‘True’ (ARa)

Misclassified (DCb)

Odds ratiod

p-value

Odds ratiod

p-value

2 × 2 tablec

0.708

0.280

0.700

0.272

logistic regression

0.360

0.326

0.530

0.540

  1. a Autopsy reports, which represent ‘true’ causes of death.
  2. b Death certificates, which represent misclassified causes of death.
  3. c The chi-squared test was used to calculate p-values.
  4. d The odds ratios for cancer among USTUR Registrants were less than 1. This was not due to a protective relationship between cancer and radiation dose. Rather, it was largely a consequence of biases in recruitment methods.
Back to article page