Table 3.

Sensitivity analysis.

Exposure	Sensitivity analysis 1		Sensitivity analysis 2		Sensitivity analysis 3
	OR	95% CI	OR	95% CI	OR	95% CI
OBS (continious)	0.96	0.93, 0.99	0.96	0.93, 1.00	0.96	0.94,0.98
OBS (tertiles)
Q1 [3, 16]	Reference		Reference		Reference
Q2 (16,24]	0.66	0.47, 0.91	0.8	0.47, 1.34	0.72	0.52, 0.98
Q3 (24,36]	0.56	0.34, 0.91	0.48	0.24, 0.94	0.60	0.41, 0.86
P for trend	0.039		0.017		< 0.05*

Sensitivity analysis 1: The model was adjusted for age, race, poverty income ratio, education and marital status, hypertension, diabetes, cardiovascular disease, hypercholesterolemia, chronic kidney disease and HEI-2015 score; Sensitivity analysis 2: Having ED was redefined to those who responded “never able”. The model was adjusted for age, race, poverty income ratio, education and marital status, hypertension, diabetes, cardiovascular disease, hypercholesterolemia, chronic kidney disease and HEI-2015 score. Sensitivity analysis 3: Missing data in covariables were multiply imputed using a multilevel approach designed for survey data. 5 imputed datasets were generated and used for regression analysis. The pooled regression coefficients were calculated based on Rubin rules. The model was adjusted for age, race, poverty income ratio, education and marital status, hypertension, diabetes, cardiovascular disease, hypercholesterolemia, chronic kidney disease.

*Specific pooled P-value is not available.