Incremental value of hormonal therapy for deep vein thrombosis prediction: an adjusted Wells score for women.

Deep venous thrombosis (DVT) management includes prediction rule evaluation to define standard pretest DVT probabilities in symptomatic patients. The aim of this study was to evaluate the incremental usefulness of hormonal therapy to the Wells prediction rules for DVT in women. We studied women undertaking compressive ultrasound scanning for suspected DVT. We adjusted the Wells score for DVT, taking into account the ß-coefficients of the logistic regression model. Data discrimination was evaluated by the receiver operating characteristic (ROC) curve. The adjusted score calibration was assessed graphically and by the Hosmer-Lemeshow test. Reclassification tables and the net reclassification index were used for the adjusted score comparison with the Wells score for DVT. We observed 461 women including 103 DVT events. The mean age was 56 years (±21 years). The adjusted logistic regression model included hormonal therapy and six Wells prediction rules for DVT. The adjusted score weights ranged from -4 to 4. Hosmer-Lemeshow test showed a nonsignificant P value (0.69) and the calibration graph showed no differences between the expected and the observed values. The area under the ROC curve was 0.92 [95% confidence interval (CI) 0.90-0.95] for the adjusted model and 0.87 (95% CI 0.84-0.91) for the Wells score for DVT (Delong test, P value < 0.01). Net reclassification index for the adjusted score was 0.22 (95% CI 0.11-0.33, P value < 0.01). Our results suggest an incremental usefulness of hormonal therapy as an independent DVT prediction rule in women compared with the Wells score for DVT. The adjusted score must be evaluated in different populations before clinical use.

Impacto da hormonioterapia associada ao escore de predição de Wells no diagnóstico de trombose venosa profunda em mulheres submetidas a ecografia vascular / Impact of hormone therapy linked to wells prediction score in the diagnosis of deep vein thrombosis in women submitted to vascular ultrasound

Introdução: A trombose venosa profunda (TVP) apresenta elevada morbimortalidade. O escore de Wells foi elaborado para melhorar a capacidade diagnóstica pré-teste para TVP. O objetivo deste estudo foi ajustar o escore de Wells para pacientes brasileiras e incluir a variável hormonioterapia (HT), comparando acurácia e poder de reclassificação do novo escore com o original de Wells. Método: Estudo observacional transversal em que foi realizada regressão logística para incluir a variável hormonioterapia (HT) ao Escore de Wells, criando um novo escore (escore HT), que foi calibrado e ajustado para a população estudada. A qualidade dos dados foi avaliada pela estatística Kappa. Resultados: Foram estudadas 461 pacientes com idade de 56,1 ± 20,8, das quais 103 tiveram diagnóstico ecográfico de TVP. O escore HT incluiu sete variáveis, pacientes que obtiveram pontuação de -4 a 0 foram consideradas de baixo risco; de 1 a 3, moderado risco; e de 4 a 11, alto risco para TVP, com calibração adequada (valor p = 0,59). A área sob a curva ROC para o escore HT foi 0,92 (IC 95% 0,90 - 0,95) e para o escore de Wells foi de 0,87 (IC 95% 0,84 – 0,91), mostrando diferença estatisticamente significativa (p < 0,05).Conclusão: A inclusão da hormonioterapia a um modelo de predição clínica demonstrou maior acurácia comparativamente ao modelo de Wells.

Introduction: Deep venous thrombosis (DVT) presents high morbidity and mortality. The Wells score is designed to improve the pretest diagnosing capacity for DVT. The purpose of this study was to adjust the Wells score for Brazilian patients and include the variable hormone therapy (HT), comparing accuracy and power of reclassification of the new score with Wells’ original score.Methods: Cross-sectional observational study in which logistic regression was performed to include the variable hormone therapy (HT) to the Wells score, creating a new score (HT score), which has been calibrated and adjusted for the population studied. Data quality was evaluated by the Kappa statistics.Results: We studied 461 patients aged 56.1 ± 20.8, of which 103 had sonographic diagnosis of DVT. The HT score included seven variables: patients who achieved a score of -4 to 0 are considered low risk; 1 to 3, moderate risk; and 4 to 11, high risk for DVT, withproper calibration (p = 0.59). The area under the ROC curve for the HT score was 0.92 (95% CI 0.90 — 0.95) and for the Wells score it was 0.87 (95% CI 0.84 — 0.91), showing a statistically significant difference (p < 0.05).Conclusion: The inclusion of hormone therapy into a clinical prediction model showed higher accuracy compared to the model of Wells.