Table 3.
Modelling of the association between CK and major bleeding, stroke or death
| Univariable | OR | CI |
| CK | 3.63 | (2.34 to 5.66) |
| Sex* | 1.27 | (0.76 to 2.13) |
| Age | 1.07 | (1.04 to 1.10) |
| BMI | 0.95 | (0.90 to 1.00) |
| Ancestry† | 1.33 | (0.74 to 2.37) |
| Diabetes mellitus | 0.57 | (0.27 to 1.23) |
| Hypertension | 0.96 | (0.58 to 1.60) |
| Baseline SBP | 1.00 | (0.99 to 1.01) |
| Baseline haematocrit | 0.99 | (0.91 to 1.01) |
| Baseline creatinine | 0.92 | (0.84 to 1.01) |
| rt-PA vs placebo | 1.59 | (0.95 to 2.66) |
| Invasive vs conservative | 1.53 | (0.92 to 2.57) |
| Multivar. (full) | B | SE | Wald | OR | CI |
| CK | 1.31 | 0.23 | 31.89 | 3.71 | 2.35 to 5.85 |
| Sex | 0.14 | 0.28 | 0.25 | 1.15 | 0.67 to 1.99 |
| Age | 0.08 | 0.02 | 20.28 | 1.08 | 1.05 to 1.12 |
| Ancestry | 0.39 | 0.32 | 1.49 | 1.48 | 0.79 to 2.78 |
| Diabetes mellitus | 0.57 | 0.42 | 1.84 | 1.76 | 0.78 to 4.01 |
| rt-PA vs placebo | 0.48 | 0.27 | 3.14 | 1.62 | 0.95 to 2.75 |
| Invasive vs conservative | 0.38 | 0.27 | 1.96 | 1.46 | 0.86 to 2.49 |
| Constant | − 8.70 | 1.15 | 57.57 | 0.00 | – |
| Multivar. (restricted) | |||||
| CK | 1.36 | 0.23 | 35.38 | 3.89 | 2.48 to 6.08 |
| Age‡ | 0.07 | 0.02 | 20.71 | 1.08 | 1.04 to 1.11 |
| Constant | − 7.86 | 1.06 | 55.07 | 0.00 | – |
Modelling of the association between (log) creatine kinase (CK) and the composite outcome (major bleeding, stroke and death).
Parameter units are as in table 1. Baseline platelet count, and fibrinogen and fibrin(ogen) degradation products at 50 min or 12 hours after start therapy were not associated with the composite outcome (OR 1; data not shown). Multivar., multivariable. Model parameters full (restricted) model: omnibus test of model coefficients χ2 63.1 (56.3) df 7 (2); −2 log likelihood 457 (470); Bayes information criterion 479 (476); Cox and Snell R square 0.04 (0.04); Nagelkerke R square 0.14 (0.13); Hosmer Lemeshow χ2 6.6 (10.2), df 8 (8); classification overall percentage 95.7 (95.6), n=1465 (1469) complete, unimputed cases. Bootstrapping OR was identical to the depicted outcomes (data not shown). Boldface font indicates statistical significance.
*Women vs men;.
†Non-white vs white.
‡OR age per 10 years (under the assumption of linearity) is 2.10 (1.53 to 2.87)). Diuretic use at admission did not affect the final model. Substituting (log) CKmax during admission with (log) peak CK on day 1 yielded OR 2.99 (2.07 to 4.30) for CK in the final model.
BMI, body mass index; rt-PA, recombinant tissue-type plasminogen activator; SBP, systolic blood pressure.