Table 4.
Poisson regression models to establish empirical relations with damage accrual (global SDI score)
| Univariate model | Multiple model | ||||
|---|---|---|---|---|---|
| ORa | 95% CI | Pseudo‐R 2 | OR | 95% CI | |
| Disease duration | 1·035 | 1·029−1·040 | 0·124 | 1·020 | 1·014−1·026 |
| Age | 1·036 | 1·031−1·040 | 0·230 | 1·034 | 1·029−1·040 |
| Ever smoker | 1·422 | 1·244−1·626 | 0·017 | 1·175 | 1·019−1·355 |
| Lupus nephritis | 1·232 | 1·076−1·410 | 0·003 | 1·498 | 1·289−1·742 |
| Daily prednisolone dose ≥ 7·5 mg (ongoing) | 1·325 | 1·151−1·526 | 0·008 | 1·727 | 1·485−2·008 |
| Statins (ongoing) | 1·822 | 1·488−2·230 | 0·023 | 1·249 | 1·013−1·539 |
| LA positivity | 1·521 | 1·315−1·760 | 0·171 | 1·268 | 1·092−1·471 |
| HCQ (ongoing) | 0·686 | 0·598−0·787 | 0·021 | 0·851 | 0·732−0·989 |
| Hypertension | 1·367 | 1·175−1·589 | 0·024 | ||
| Triple positivity | 1·298 | 1·058−1·593 | <0·001 | ||
| Total pseudo‐R 2 (multiple model) | 0·471 | ||||
Pseudo‐R 2 is different from the R 2 used in ordinary least‐squares regression models. However, it will give an approximation of how well the independent variables are related with the outcome [sum of global SLICC/ACR damage index (SDI)]. CI = confidence intervals.
a
The odds ratios (OR) can be interpreted as follows: an increase of 1 year of disease duration is associated with 3·5% higher score (OR = 1·035) in the number of SDI points, and ongoing treatment with hydroxychloroquine (HCQ) is associated with 31·4% less score (OR = 0·686, 1–0·686 = 0·314) in the sum of global SDI score compared to those not having ongoing treatment with HCQ.