Table 3.
Results of Logistic Regression Models on Warfarin Treatment for Atrial Fibrillation (i.e., of those self-reporting AF, what are the predictors of being on warfarin; n = 258).
| Univariate Models* | Full Multivariable Model† | Most Parsimonious Model‡ (α=0.2) | |
|---|---|---|---|
| OR (95% CI) | OR (95% CI) | ||
| Race | |||
| Black vs. White | 0.28 (0.13, 0.60) | 0.31 (0.10, 0.90) | 0.26 (0.12, 0.57) |
| Sex | |||
| Female vs. Male | 1.06 (0.55, 2.05) | 0.84 (0.34, 2.06) | |
| Region | |||
| Belt vs. Nonbelt | 0.62 (0.31,1.22) | 0.51 (0.21, 1.22) | |
| Buckle vs. Nonbelt | 1.21 (0.51, 2.89) | 1.18 (0.421, 3.29) | |
| Age | |||
| Less than 75 vs. 75 or older | 0.96 (0.52, 1.77) | 0.59 (0.26, 1.33) | |
| Education | |||
| Less than High School vs. College or Higher | 0.61 (0.21,1.80) | 0.89 (0.17, 4.75) | |
| High School vs. College or Higher | 1.61 (0.70, 3.68) | 1.87 (0.64, 5.48) | |
| Some College vs. College or Higher | 0.90 (0.43, 1.90) | 0.83 (0.32, 2.13) | |
| Income | |||
| <$20,000 vs. >$75,000 | 0.68 (0.21, 2.19) | 1.48 (0.28, 7.94) | |
| $20,000-$35,000 vs. >$75,000 | 0.85 (0.29, 2.45) | 1.18 (0.31, 4.44) | |
| $35,000-$75,000 vs. >$75,000 | 0.81 (0.29, 2.30) | 1.01 (0.31, 3.36) | |
| Refused vs. >$75,000 | 0.52 (0.17, 1.63) | 0.46 (0.11, 1.93) | |
| Insurance Status | |||
| Covered vs. Not Covered | 4.08 (0.56, 29.67) | 5.38 (0.60, 47.83) | 5.14 (0.70, 37.68) |
| Access to Medical Care | |||
| No vs. Yes | 1.29 (0.53, 3.10) | 1.34 (0.44, 4.07) | |
| Congestive Heart Failure | |||
| No vs. Yes | 2.10 (0.75, 5.88) | 1.49 (0.31, 7.12) | |
| Hypertension | |||
| No vs. Yes | 1.22 (0.64, 2.33) | 1.29 (0.57, 2.94) | |
| Diabetes | |||
| No vs. Yes | 0.98 (0.48, 2.03) | 0.66 (0.25, 1.72) | |
| Prior Stroke or TIA | |||
| No vs. Yes | 1.04 (0.49, 2.20) | 1.05 (0.26, 4.33) | |
| Presence of Stroke Symptoms | |||
| No vs. Yes | 1.01 (0.44, 2.28) | 0.63 (0.21, 1.89) |
*
Univariate column of the table is the isolated association of the specific factor with the outcome of treatment with warfarin.
†
Multivariable column is the association of the specific factor with the outcome after adjustment for all other factors shown in the table.
‡
Parsimonious column is the multivariable association of those factors that remained significant with the particular outcome after adjustment for the other factors that remained significant.