Computer algorithm compared with routine clinical practice in screening for deep venous thrombosis in the leg in an emergency department

Abstract

ObjectiveTo audit routine clinical practice in screening patients admitted to an accident and emergency department for deep venous thrombosis in the leg (DVT) and compare it with a computer algorithm. DesignRetrospective study of case notes with data subsequently entered into a computer algorithm followed by a comparison of the diagnosis and recommended management from each modality. SettingEmergency department at a London teaching hospital. Population or SampleA convenience sample of 43 patients attending the emergency department. MethodsEvaluation of clinical notes for completeness of assessment and correct diagnosis; entry of the same data into the computer algorithm. Main Outcome MeasuresCompleteness of data collection, diagnosis of presence or absence of DVT, and recommended therapy. ResultsThe Wells score was missing in the clinical evaluation in 60% of cases. Clinicians relied primarily on the results of ultrasound scans and in six cases the absence of required D-Dimer measurements meant that the algorithm stalled. Clinical and algorithm analysis and recommendations were not compliant in 10 cases (23%). ConclusionsClinical assessment of potential DVT in the accident and emergency department is poorly performed when compared with a computerised algorithm. Clinicians rely heavily on scan reports rather than clinical assessment which may have cost implications.