Venous Thrombosis Risk after Cast Immobilization of the Lower Extremity: Derivation and Validation of a Clinical Prediction Score, L-TRiP(cast), in Three Population-Based Case–Control Studies

Background: Guidelines and clinical practice vary considerably with respect to thrombosis prophylaxis during plaster cast immobilization of the lower extremity. Identifying patients at high risk for the development of venous thromboembolism (VTE) would provide a basis for considering individual thromboprophylaxis use and planning treatment studies. Methods and Findings: We used data from a large population-based case–control study (MEGA study, 4,446 cases with VTE, 6,118 controls without) designed to identify risk factors for a first VTE. Cases were recruited from six anticoagulation clinics in the Netherlands between 1999 and 2004; controls were their partners or individuals identified via random digit dialing. Identification of predictor variables to be included in the model was based on reported associations in the literature or on a relative risk (odds ratio) > 1.2 and p ≤ 0.25 in the univariate analysis of all participants. Using multivariate logistic regression, a full prediction model was created. In addition to the full model (all variables), a restricted model (minimum number of predictors with a maximum predictive value) and a clinical model (environmental risk factors only, no blood draw or assays required) were created. To determine the discriminatory power in patients with cast immobilization (n = 230), the area under the curve (AUC) was calculated by means of a receiver operating characteristic. Validation was performed in two other case–control studies of the etiology of VTE: (1) the THE-VTE study, a two-center, population-based case–control study (conducted in Leiden, the Netherlands, and Cambridge, United Kingdom) with 784 cases and 523 controls included between March 2003 and December 2008 and (2) the Milan study, a population-based case–control study with 2,117 cases and 2,088 controls selected between December 1993 and December 2010 at the Thrombosis Center, Fondazione IRCCS Ca’ Granda–Ospedale Maggiore Policlinico, Milan, Italy. Conclusions: These results show that information on environmental risk factors, coagulation factors, and genetic determinants in patients with plaster casts leads to high accuracy in the prediction of VTE risk. In daily practice, the clinical model may be the preferred model as its factors are most easy to determine, while the model still has good predictive performance. These results may provide guidance for thromboprophylaxis and form the basis for a management study. Using three population-based case-control studies, Banne Nemeth and colleagues derive and validate a clinical prediction score (L-TRiP(cast)) for venous thrombosis risk.Background: Blood normally flows smoothly around the human body, but when a cut or other injury occurs, proteins called clotting factors make the blood gel (coagulate) at the injury site. The resultant clot (thrombus) plugs the wound and prevents blood loss. Sometimes, however, a thrombus forms inside an uninjured blood vessel and partly or completely blocks the blood flow. Clot formation inside one of the veins deep in the body (usually in a leg) is called deep vein thrombosis (DVT). DVT, which can cause pain, swelling, and redness in the affected limb, is treated with anticoagulants, drugs that stop the clot growing. If left untreated, part of the clot can break off and travel to the lungs, where it can cause a life-threatening pulmonary embolism. DVT and pulmonary embolism are known collectively as venous thromboembolism (VTE). Risk factors for VTE include age, oral contraceptive use, having an inherited blood clotting disorder, and prolonged inactivity (for example, being bedridden). An individual’s lifetime risk of developing VTE is about 11%; 10%–30% of people die within 28 days of diagnosis of VTE. Why Was This Study Done?: Clinicians cannot currently accurately predict who will develop VTE, but it would be very helpful to be able to identify individuals at high risk for VTE because the condition can be prevented by giving anticoagulants before a clot forms (thromboprophylaxis). The ability to predict VTE would be particularly useful in patients who have had a lower limb immobilized in a cast after, for example, breaking a bone. These patients have an increased risk of VTE compared to patients without cast immobilization. However, their absolute risk of VTE is not high enough to justify giving everyone with a leg cast thromboprophylaxis because this therapy increases the risk of major bleeds. Here, the researchers investigate the predictive value of genetic and environmental factors and levels of coagulation factors and other biomarkers on VTE occurrence after cast immobilization of the lower leg and develop a clinical tool for the prediction of VTE in patients with plaster casts. What Did the Researchers Do and Find?: The researchers used data from the MEGA study, a study of risk factors for VTE, to build a prediction model for a first VTE in patients with a leg cast; the prediction model included 32 predictors (the full model). They also built a restricted model, which included only 11 predictors but had maximum predictive value, and a clinical model, which included 14 environmental predictors that can all be determined without drawing blood or undertaking any assays. They then determined the ability of each model to distinguish between patients with a leg cast who did and did not develop VTE using receiver operating characteristic (ROC) curve analysis. The area under the curve (AUC) for the full model was 0.85, for the restricted model it was 0.85, and for the clinical model it was 0.77. (A predictive test that discriminates perfectly between individuals who do and do not subsequently develop a specific condition has an AUC of 1.00; a test that is no better at predicting outcomes than flipping a coin has an AUC of 0.5.) Similar or higher AUCs were obtained for all the models using data collected in two independent studies. Finally, the researchers converted the clinical model into a risk score by giving each variable in the model a numerical score. The sum of these scores was used to stratify individuals into categories of low or high risk for VTE. With a cutoff of 9 points, the risk score correctly identified 80.8% of the patients in the MEGA study with a plaster cast who developed VTE and 60.8% of the patients who did not develop VTE. What Do These Findings Mean?: Some aspects of this study may limit the accuracy of its findings. For example, no information was available about which patients with a plaster cast received thromboprophylaxis. Nevertheless, these findings suggest that information on environmental risk factors, coagulation factors, and genetic determinants can be used to predict VTE risk in patients with a leg cast with high accuracy. Importantly, the risk score derived and validated by the researchers, which includes only predictors that can be easily determined in clinical practice, may help clinicians decide which patients with a leg cast should receive thromboprophylaxis and which should not be exposed to the risk of anticoagulant therapy, until an unambiguous guideline for these patients becomes available. Additional Information: This list of resources contains links that can be accessed when viewing the PDF on a device or via the online version of the article at http://dx.doi.org/10.1371/journal.pmed.1001899.