Research Article

First Use of Multiple Imputation with the National Tuberculosis Surveillance System

Table 2

Logistic regression models for death: multiple imputation versus case exclusion.

TermMultiple imputationCase exclusion

Isoniazid resistance2.07 (1.30, 3.29)1.53 (0.83, 2.83)
Male gender1.20 (0.88, 1.63)1.16 (0.81, 1.66)
Age categories
 Age ≤ 10.076 (0.008, 0.69) +
 1 < Age ≤ 40.23 (0.067, 0.82)0.61 (0.13, 2.79)
 4 < Age ≤ 140.38 (0.102, 1.41) +
 14 < Age ≤ 241.22 (0.64, 2.34)1.44 (0.71, 2.93)
 24 < Age ≤ 34Reference
 34 < Age ≤ 441.30 (0.87, 1.92)1.11 (0.69, 1.78)
 44 < Age ≤ 541.97 (1.23, 3.15)1.87 (1.12, 3.13)
 54 < Age ≤ 641.83 (1.09, 3.09)1.13 (0.57, 2.23)
 64 < Age ≤ 744.36 (2.48, 7.67)5.85 (2.80, 12.19)
 Age > 746.90 (3.85, 12.38)2.99 (1.25, 7.12)
Race categories
 WhiteReference
 Black1.44 (1.01, 2.06)1.10 (0.69, 1.76)
 Hispanic1.21 (0.74, 1.99)1.19 (0.71, 1.98)
 Asian*0.64 (0.30, 1.35)0.62 (0.30, 1.28)
 American Indian9.07 (2.65, 31.02)28.5 (3.12, 260.2)
 Other**0.85 (0.20, 3.52)0.81 (0.15, 4.43)
HIV positive3.57 (1.87, 6.82)4.58 (3.13, 6.71)

*Includes Native Hawaiian.
**Includes multiple race and unknown categories.
+There were no deaths among patients with known HIV status in this age group.