Skip to main content

Table 4 Multinomial logistic regression model for redistribution of X59 deaths

From: Injury death certificates without specification of the circumstances leading to the fatal injury – the Norwegian Cause of Death Registry 2005–2014

Complete model

Accuracy (mean(SD))

Kappa (mean(SD))

LR stat (mean(SD))*

p value

0.712 (0.01)

0.636 (0.01)

Ref.

 

Reduced models

Without (one at a time)

  Nature of injury

0.494 (0.01)

0.341 (0.01)

10,582 (147)

<  0.001

  Place of death

0.673 (0.01)

0.586 (0.01)

1563 (61)

<  0.001

  Age

0.690 (0.01)

0.607 (0.01)

1298 (56)

<  0.001

  Autopsy

0.706 (0.01)

0.629 (0.01)

366 (26)

<  0.001

  Scene of injury

0.704 (0.01)

0.627 (0.01)

309 (27)

<  0.001

  Gender

0.712 (0.01)

0.636 (0.01)

199 (22)

<  0.001

  Calendar year of death

0.711 (0.01)

0.635 (0.01)

33 (8)

<  0.001

  1. Data from the Norwegian Cause of Death Registry, 2005–2014
  2. Based on 1000 repetitions of random division into new training and test sets. The models were developed on the training sets and evaluated on the test sets
  3. *The likelihood ratio statistic (−2 logL) is computed by comparing the full model to the model without the variable in question. The higher the LR statistics, the more the model is weakened by excluding the variable in question