ABSTRACT
INTRODUCTION: Routine testing for thrombophilia following venous thromboembolism (VTE) is controversial. The use of large datasets to study the clinical impact of thrombophilia testing on patterns of care and patient outcomes may enable more efficient analysis of this practice in a wide range of settings. We set out to examine how accurately algorithms using International Classification of Diseases 9th Revision (ICD-9) codes and/or pharmacy data reflect laboratory-confirmed thrombophilia diagnoses. MATERIALS AND METHODS: A random sample of adult Kaiser Permanente Colorado patients diagnosed with unprovoked VTE between 1/2004 and 12/2010 underwent medical record abstraction of thrombophilia test results. Algorithms using "ICD-9" (positive if a thrombophilia ICD-9 code was present), "Extended anticoagulation (AC)" (positive if AC therapy duration was >6 months), and "ICD-9 & Extended AC" (positive for both) criteria to identify possible thrombophilia cases were tested. Using positive thrombophilia laboratory results as the gold standard, the sensitivity, specificity, positive predictive value (PPV), and negative predictive value of each algorithm were calculated, along with 95% confidence intervals (CIs). RESULTS: In our cohort of 636 patients, sensitivities were low (<50%) for each algorithm. "ICD-9" yielded the highest PPV (41.5%, 95% CI 26.3-57.9%) and a high specificity (95.9%, 95% CI 94.0-97.4%). "Extended AC" had the highest sensitivity but lowest specificity, and "ICD-9 & Extended AC" had the highest specificity but lowest sensitivity. CONCLUSIONS: ICD-9 codes for thrombophilia are highly specific for laboratory-confirmed cases, but all algorithms had low sensitivities. Further development of methods to identify thrombophilia patients in large datasets is warranted.
Subject(s)
Algorithms , Blood Coagulation Tests/statistics & numerical data , International Classification of Diseases/statistics & numerical data , Thrombophilia/diagnosis , Thrombophilia/prevention & control , Venous Thromboembolism/therapy , Adult , Aged , Aged, 80 and over , Anticoagulants/administration & dosage , Colorado/epidemiology , Comorbidity , Decision Support Systems, Clinical , Diagnosis, Computer-Assisted/methods , Female , Humans , Incidence , Male , Middle Aged , Prevalence , Reproducibility of Results , Risk Assessment/methods , Sensitivity and Specificity , Thrombophilia/epidemiology , Venous Thromboembolism/diagnosis , Venous Thromboembolism/epidemiologyABSTRACT
BACKGROUND: Thrombophilia testing has limited value in determining the selection and duration of anticoagulation therapy for venous thromboembolism (VTE), yet is commonly performed. This study describes the patterns and appropriateness of thrombophilia testing in a large cohort of patients with acute VTE. MATERIALS AND METHODS: This was a retrospective study of a random sample of patients with a validated diagnosis of acute VTE diagnosed between January 1, 2004 and December 31, 2010. Events were identified from administrative data and verified via manual review. Patients were grouped by thrombophilia testing status and compared on patient characteristics and thrombophilia testing results and appropriateness. RESULTS: Of 1314 patients with validated VTE, 315 (24%) underwent thrombophilia testing, 62 (20%) of whom had ≥ 1 positive test. Tested patients were younger and more likely to have had a family history of VTE. Factor V Leiden (17%) and prothrombin G20210A mutation (4%) were the most commonly detected thrombophilias. Only 31 (10%) of tested patients met eligibility criteria for thrombophilia testing (i.e., at least one strong thrombophilic risk factor present) and were tested at least 90 days following unprovoked index VTE. CONCLUSIONS: Thrombophilia is commonly evaluated in patients without a clear indication for testing and during times where results may be unreliable. Future studies are needed to assess interventions aimed at reducing inappropriate thrombophilia testing without adversely affecting patient outcomes.
