Predictors of anticoagulation quality in 15’834 patients performing
patient self-management of oral anticoagulation with vitamin K
antagonists in real-life practice: a survey of the International SelfMonitoring Association of Oral Anticoagulated Patients
Christian Schaefer1, Michael Nagler2,3, MD, PhD; Walter A. Wuillemin2,4, MD, PhD
1International Self-Monitoring Association of Oral Anticoagulated Patients (ISMAAP), Geneva, Switzerland; 2Division of Hematology
and Central Hematology Laboratory, Luzerner Kantonsspital, CH-6000 Lucerne 16, Switzerland; 3Laboratory of Clinical Thrombosis
and Haemostasis, and Cardiovascular Research Institute, Maastricht University Medical Center, 6229 ER Maastricht, The
Netherlands; 4University of Berne, CH-3010 Berne
Background
Figure 1: Characteristics of patients performing PSM in real-life
clinical practice
Patient self-management (PSM) of oral anticoagulation with vitamin
K antagonists (VKA) is recommended for patients requiring longterm anticoagulation but some aspects are still under debate. It is
not known how to identify patients with a low chance of performing
well.
Methods
Data from a large survey of the International Self-Monitoring
Association of Orally Anticoagulated Patients (ISMAAP) were used.
The percentages of INR values within therapeutic range (TIR) were
calculated as outcome variable, and various characteristics were
obtained as potential predictors. Logistic regression for low
anticoagulation control (TIR below 75%) was used. A prediction
model was fitted by comparing receiver-operating characteristics
(ROC curves).
Results
Table 1: Predictors of low anticoagulation control (univariate
analysis)
Median age was 72 years (IQR 65 to 77 years), 30.1% were female.
Indications for anticoagulation were as follows: mechanical heart
valve (46.5%), atrial fibrillation (34.3%), venous thromboembolism
(16.4%) and others (2.9%). The median TIR was 88.5% (IQR 76.9 to
96.2%). The patient characteristics are shown in Figure 1, predictors
of anticoagulation quality in Table 1 and the final prediction model in
Table 2.
Table 2: Prediction model for low anticoagulation quality*
* TIR below 75%; # compared to the previous model; + final
model, sensitivity (threshold 0.2): 85.3% (95%CI: 84.0, 86.4),
specificity: 28.5% (95%CI: 27.6, 29.5)
Discussion
Using the proposed prediction model, physicians will be able to
identify patients with a low chance of performing well, considering
additional teaching lessons, regular follow-up, or adjustment of
therapeutic ranges. Intensifying support in this patient group will
improve care in PSM patients.