Table 2.

Polytomous logistic regression comparing endorsement of one item indicating alcohol misuse and of two or more items indicating alcohol misuse with endorsement of no items indicating misuse.

	One alcohol misuse item endorsed	Two or more alcohol misuse items endorsed
	OR (99% CI) [P value]	OR (99% CI) [P value]
Sociodemographic characteristics
Sex: Male	8.19 (5.56, 12.06) [0.001]	20.81(13.62, 31.79) [0.001]
Age:
60–69	Reference	Reference
70–79	0.81 (0.57, 1.15) [0.12]	0.69 (0.49, 0.96) [0.004]
80+	0.57 (0.29, 1.10) [0.03]	0.42 (0.22, 0.79) [0.001]
Income: Lower income	1.19 (0.85, 1.67) [0.18]	1.42 (1.03, 1.95) [0.005]
Race:
White	Reference	Reference
African-Brazilian	1.17 (0.62, 2.22) [0.09]	2.85 (1.78, 4.57) [0.001]
Other	1.36 (0.85, 2.17) [0.60]	1.34 (1.85, 2.10) [0.001]
Lifestyle & social environment
Tobacco use: Yes	2.02 (1.43, 2.83) [0.001]	2.44(1.79,3.32) [0.001]
Marital status:
No longer married	Reference	Reference
Married	0.91 (0.63, 1.32) [0.53]	0.77 (0.55, 1.08) [0.05]
Never married	1.61 (0.87, 2.96) [0.05]	1.49 (0.84, 2.65) [0.07]
Health status and health conditions
Impaired activities of daily living:
0 problems	Reference	Reference
1–2 problem	1.26 (0.88, 1.80) [0.10]	1.12 (0.79, 1.59) [0.39]
3+ problems	1.14 (0.58, 2.24) [0.61]	1.93 (1.13, 3.30) [0.001]
Vascular condition - Yes	0.74 (0.53, 1.03) [0.02]	0.63 (0.46, 0.86) [0.001]
Respiratory condition - Yes	1.22 (0.87, 1.71) [0.12]	1.43 (1.05, 1.94) [0.003]
Psychiatric problem - Yes	1.57 (1.11, 2.22) [0.001]	2.19 (1.58, 3.03) [0.001]

The variables considered are the sociodemographic, lifestyle and social environment, and health conditions listed in Table 1. The significant variables entered into a prior analysis (which required P>0.05 to enter) were selected for this final model.

OR = Odds Ratio; CI = Confidence Interval