Table 2.

 Unadjusted differences in proportions and multivariable analyses of associations with likelihood of choosing comfort care as primary goal of care

Characteristics	Frequency choosing comfort care	Difference in % choosing comfort care (95% CI)	Unadjusted P value	Odds ratio (95% CI)
				Unadjusted	Adjusted
Age (years):
<80†	104 (71)	12% (−0.3% to 25%)	0.10	2.0 (0.9 to 4.5)	—
≥80	45 (83)
Sex:
Female†	88 (76)	−3% (−16% to 9%)	0.62	0.8 (0.4 to 1.6)	—
Male	61 (73)
Education:
<College graduate†	79 (68)	15% (3% to 27%)	0.021	2.3 (1.2 to 4.6)	—
≥College graduate	69 (83)
Marital status:
Not married†	82 (71)	8% (−4% to 20%)	0.25	1.5 (0.8 to 2.9)	—
Married	67 (79)
Health status‡:
Fair or poor†	35 (58)	23% (9% to 37%)	0.001	3.0 (1.6 to 6.0)	—
Good or better	111 (81)
Diagnosis of dementia:
No†	136 (75)	−3% (−25% to 19%)	1.0	0.9 (0.3 to 2.5)	—
Yes	13 (72)
Previous relationship with person with advanced dementia:
No†	125 (73)	10% (−6% to 25%)	0.36	1.8 (0.6 to 4.9)	—
Yes	24 (83)
Randomisation:
Verbal†	68 (64)	22% (11% to 34%)	<0.001	3.5 (1.7 to 7.1)	3.9 (1.8 to 8.6)
Video	81 (86)
Health literacy§:
≤6th grade†	16 (46)
7-8th grades	14 (58)	13% (−13% to 38%)		1.7 (0.6 to 4.7)	1.7 (0.54 to 5.3)
≥9th grade	119 (84)	39% (21% to 56%)	<0.001	6.4 (2.9 to 14.4)	4.1 (1.6 to 10.8)
Race:
Black/ African-American†	30 (51)	34% (19% to 48%)	<0.001	5.2 (2.6 to 10.4)	2.9 (1.3 to 6.6)
White	119 (84)

*For multivariable analysis, characteristics excluded from model if they were not related to outcome—that is, choosing comfort care—at P<0.05.

†Reference category.

‡Health status was one of excellent, very good, good, fair, or poor.

§Assessed with rapid estimate of adult literacy in medicine (REALM).