Skip to main content
. Author manuscript; available in PMC: 2015 Sep 20.
Published in final edited form as: Stat Med. 2014 May 7;33(21):3710–3724. doi: 10.1002/sim.6173

Table VI.

Two post-imputation analysis results from alternative missing data methods using the medical records and medicare claims, a subsample of CanCORS data.

Variables MI OR AND Medical
records
Medicare
claims

Hospice-use 62.8% 67.6% 39.0% 29.3% 55.4%
65–69 years .56* −1.51* −.05 −.11 NA
70–74 years .26 −1.61* .08 −.03 .10
75–79 years .36 −1.52* .13 .04 .09
80 years or older .68* −1.22* .26 .08 .42*
Female .12 .27* .20 −.03 .24*
Black −.31 −.28 −.69* −.89* −.20
Hispanic .31 .11 −.63 −.55* −.17
Asian −.36 −.42 −1.54* −.62* −.66*
Other race −.58 −.58* −1.12* −.50* −.83*
Less than high school .11 .00 .33 .33* .02
High school −.03 −.10 .25 .42* −.17
Less than 20K .23 −.16 −.08 .01 .18
20–39K .32 −.09 −.18 −.05 .22
40–59K .44* .14 −.10 .12 .37
Married −.23 −.16 −.05 .02 −.14
Stage 2 .49 .34 .20 −.04 .30
Stage 3 .48* .35 .23 .13 .29
Stage 4 .90* .48* .21 .21 .34
Stage missing .78* .68* −.10 −.22 .44
MI −.02 −.18 −.16 −.02 −.11
CHF −.27 −.04 −.04 .06 −.06
Stroke .08 −.09 .04 .17 −.10
Lung disease −.13 −.11 −.10 −.18* −.03
Diabetes .03 .09 .30* .04 .28*
Depression .33* .18 .03 .00 .15

Note: The subsample consists of 2669 CanCORS patients who died within 15 months of diagnosis and had at least one observed hospice-use record from two sources. The first row (‘hospice use’) lists marginal estimates of the hospice-use rate. Following rows list the logistic regression coefficients (Section 3.4.2). MI, multiple imputation. Standard errors are not shown. Logistic regression coefficients that are significant at 10% level are highlighted. The reference categories are lung cancer, younger than 65 years old for MI, OR, AND, and Medical records/younger than 70 years old for Medicare Claims, White, college, more than 60K, stage I.