Table 2.
Ordinal regression models of association between BMI (independent variable) and HAQ score (dependent variable) after imputation of missing data*
| Univariate | Multivariate | |||
|---|---|---|---|---|
| Independent variable | OR (95% CI) | FMI | OR (95% CI) | FMI |
| BMI (kg/m2) | ||||
| <25 | Referent | Referent | ||
| 25≤ BMI <30 | 0.92 (0.72–1.17) | 0.03 | 1.04 (0.79–1.35) | 0.05 |
| 30≤ BMI <35 | 1.32 (0.98–1.77) | 0.03 | 1.15 (0.82–1.61) | 0.05 |
| ≥35 | 2.32 (1.49–3.61)† | 0.02 | 1.94 (1.17–3.24)† | 0.08 |
| Age, years | – | – | 1.03 (1.02–1.04)† | 0.13 |
| Female sex | – | – | 1.88 (1.46–2.41)† | 0.05 |
| DAS28 | – | – | 2.44 (2.16–2.75)† | 0.24 |
| Current smoker (vs. nonsmoker) | – | – | 1.69 (1.23–2.34)† | 0.09 |
| On treatment | – | – | 1.53 (1.20–1.95)† | 0.09 |
*
BMI = body mass index; HAQ = Health Assessment Questionnaire; OR = odds ratio; 95% CI = 95% confidence interval; FMI = fraction of missing information; DAS28 = Disease Activity Score in 28 joints.
†Significant at P < 0.05.