Table 1. Stepwise Logistic Regression Models of the Association of Self-Reported Hearing Impairment and Frailty in Adults aged 70 and Older in NHANES 1999–2002.

Self-reported hearing impairment was defined as good/little trouble hearing versus a lot of trouble hearing. Frailty was classified as being not frail versus pre-frail/frail.^a

	All Participants		Men		Women
	Odds Ratio of Frailty (95% CI)	p-value	Odds Ratio of Frailty (95% CI)	p-value	Odds Ratio of Frailty (95% CI)	p-value
Hearing loss alone	1.79 (1.29, 2.50)	.001	1.29 (0.83, 2.01)	.25	3.47 (1.97, 6.10)	<.001
Base model (Hearing loss + age)	1.52 (1.08, 2.14)	.02	1.20 (0.74, 1.93)	.45	2.74 (1.46, 5.12)	.003
Base model + demographic factorsb	1.59 (1.12, 2.25)	.01	0.95 (0.55, 1.65)	.86	2.78 (1.56, 4.97)	.001
Base model + demographic factors + cardiovascular risk factorsc	1.54 (1.05, 2.25)	.03	0.83 (0.47, 1.46)	.51	2.84 (1.55, 5.20)	.001
Base model + demographic factors + cardiovascular risk factors + hearing aid use	1.80 (1.08, 2.98)	.02	0.95 (0.50, 1.81)	.88	3.60 (1.74, 7.44)	.001
Base model + demographic factors + cardiovascular risk factors + hearing aid use + general health status	1.68 (1.00, 2.82)	.05	0.85 (0.44, 1.66)	.62	3.79 (1.69, 8.51)	.002

^aSampling weights were used in these calculations.

^bDemographic factors include: gender, race, education, income, and body mass index (BMI). Note gender was removed from analyses stratified by gender.

^cCardiovascular risk factors include: smoking status, hypertension, stroke, and diabetes.