Cardiovascular risk factors are major determinants of thrombotic risk in patients with the lupus anticoagulant

Table 4 Multivariable models for thrombotic risk in patients with LA

Models	SHR	95% CI	p	Log(SHR)	Assigned point
Model 1
Diabetes	3.97	1.29-12.19	0.016	1.38	2
Active smoking	2.42	1.15-5.06	0.019	0.88	1
Prolonged aPTT-LA ratio^a	2.28	1.04-4.99	0.039	0.82	1
Model 2
0 point (n = 77 (51.3%))	Ref.	Ref.	Ref.	-	-
1 point (n = 51 (34.0%))	2.84	1.14-7.02	0.024	-	-
2 points (n = 15 (10.0%))	8.56	2.91-25.17	<0.0001	-	-
3 points (n = 7 (4.7%))	8.45	2.21-32.35	0.002	-	-
Model 3
0 point (n = 77 (51.3%))	Ref.	Ref.	Ref.	-	-
1 point (n = 51 (34.0%))	2.84	1.14-7.02	0.024	-	-
≥ 2 points (n = 22 (14.7%))	8.53	3.19-22.78	<0.0001	-	-

Model 1 is a multivariable model including the three variables as binary specifications (a prolonged lupus-sensitive aPTT was defined as being above the 75th percentile (Q3) of this variable’s distribution (cut-off: 117.5 s)). Model 2 is a multivariable model based on the points that were assigned for the relative contribution of the individual variables (as represented by the log (subdistribution hazard ratios) in Model 1). In Model 3, the two highest risk categories were combined in a post hoc fashion, because the coefficients showed a similar relative risk for 2 and 3 points. Model 3 is the final product of our prediction model-building strategy, and observed risks according to this point-based rule are shown in Fig. 1
SHR subdistribution hazard ratio, 95%CI 95% confidence interval, p Wald test p value, log(SHR) natural logarithm of the SHR, Ref reference category
^aProlonged aPTT ratio defined by a prolongation above the 75th percentile of this variable’s distribution (this corresponds to cut-off at 117.5 s (or for the aPTT ratio at 3.4 multiples of the median in healthy individuals))

ISSN: 1741-7015