Table 5.

Associations Between Patients' Self-Estimates of LE, EOL Care Preferences, and Advance Care Planning (n = 299)

Outcome	Self-Estimate of LE (months)	Unadjusted Analysis		Adjusted Analysis
	Median (Q1 to Q3)	Corrected OR (95% CI)*	P	Corrected OR (95% CI)	P
Prefers life-prolonging care†		1.638 (1.241 to 2.064)	< .001	1.493 (1.091 to 1.939)	.0138
Yes (n = 71)	60.0 (12.0 to 240.0)
No (n = 206)	24.0 (12.0 to 120.0)
DNR order‡		0.406 (0.280 to 0.575)	< .001	0.439 (0.296 to 0.630)	< .001
Yes (n = 117)	13.0 (6.0 to 60.0)
No (n = 174)	60.0 (15.0 to 240.0)
Living will/health care proxy§		0.526 (0.386 to 0.686)	< .001	0.722 (0.539 to 0.906)	.0024
Yes (n = 199)	24.0 (8.0 to 60.0)
No (n = 92)	97.0 (18.0 to 240.0)

Abbreviations: DNR, do not resuscitate; EOL, end of life; LE, life expectancy; OR, odds ratio; Q, quartile.

^*For all analyses: because independent variable (patients' self-estimated prognosis in months) was not normally distributed, this variable was transformed by log base 12 for purposes of analysis; thus, ORs represent change in odds of outcome (eg, DNR order) for every 12-month increase in patients' self-estimated prognosis.

^†ORs have been corrected using method of Zhang et al⁵² for correcting ORs in cohort studies of common outcomes.

^‡Adjusted for age, sex, and recruitment site.

^§Adjusted for sex and Eastern Cooperative Oncology Group performance status.

^‖Adjusted for education, white race, Baptist religion, and recruitment site.