Investigation of two general venous thromboembolism risk-stratification models in predicting venous thromboembolic events in TB patients.

Background: Tuberculosis (TB) remains an infectious disease with a high prevalence worldwide and represents a major public health concern. It is known that TB causes a hyper-coagulable state due to its infective nature. Thus, patients are prone to higher incidence rates of venous thromboembolism (VTE) in comparison with the general population. Although there are many risk assessment models (RAMs) that estimate the risk of VTE incidence in the general population, none have been proven to show significant prognostic value in early and accurate VTE prediction among TB patients. This study aims to investigate the accuracy of general RAMs in prediction of VTE incidence in the population of TB-positive patients.

Methods: The following survey is a retrospective study among patients afflicted with TB, in whom VTE had occurred either during or at the onset of admission. A total number of 865 smear-positive TB cases were recorded in hospital in a 7-year timespan, among which a total sum of 37 patients (67.6% males, age: 56.19 ± 20 years) experienced at least one episode of thromboembolic incidence. We nominated Padua and Geneva RAMs and calculated the scores with regard to their inpatient hospital records.

Results: Of 865 adult hospitalized smear-positive TB patients, 37 patients happened to develop VTE in the course of infection, after excluding the unacceptable data. The incidence of VTE was calculated at 4.27%. Of all VTEs, 73% turned out to be deep venous thromboembolism (DVT), 18.9% were pulmonary thromboembolism (PTE), and 8.1% of patients developed both DVT and PTE during the course of disease. Among all cases, 32.4% revealed Geneva score equivalent or >3. It conveys the meaning that these patients were at greater risk for VTE development and were indicated to receive prophylactic medication. Similarly, Padua model was capable of predicting 29.7% cases scoring >4, which is alarming for elevated VTE probability. 21.6% of TB patients, who had developed VTE during the course of their disease eventually passed away.

Conclusion: Our statistics show minimal positive predictive value for Padua and Geneva RAMs, which are seemingly in sharp contrast with the excellent validation of these models verified by numerous surveys in general population. This fact could be attributable to failing to consider TB, or in general chronic infections, as independent indicators of VTE incidence. These findings indicate the need for revising the presenting RAMs or establishing a separate RAM for VTE prediction in TB patients, resembling the VTE risk assessment model for cancer patients.