Multivariable model of predictors for the development of hypogammaglobulinaemia in the RITAZAREM trial
| Variable | OR | 95% CI | P value |
| Maintenance treatment (rituximab vs azathioprine) | 2.2 | (0.87 to 5.72) | 0.104 |
| Glucocorticoid induction regimen (1A vs 1B) | 8.6 | (3.02 to 27.58) | <0.001 |
| ANCA status at enrolment (anti-PR3 vs anti-MPO) | 0.78 | (0.27 to 2.26) | 0.639 |
| Type of relapse (severe vs non-severe) | 2.2 | (0.82 to 6.07) | 0.124 |
| Previous rituximab (yes vs no) | 1.2 | (0.41 to 3.53) | 0.740 |
| Previous cyclophosphamide (yes vs no) | 1.1 | (0.20 to 6.36) | 0.944 |
| Previous rituximab or cyclophosphamide (yes vs no) | 7.3 | (0.57 to 128.37) | 0.144 |
| Baseline plasma IgG (g/L)* | 0.52 | (0.40 to 0.65) | <0.001 |
| Intercept | 0.013 | (0.00 to 0.12) | – |
Glucocorticoid induction regimen: 1A=1 mg/kg/day starting dose; 1B=0.5 mg/kg/day starting dose.
*Baseline plasma IgG level was centred around the mean value (9.56). The intercept gives an estimate of the absolute odds for a patient with the mean value of baseline IgG, and the reference levels of the other binary predictors.
MPO, myeloperoxidase; PR3, proteinase 3.