Diagnostic and prognostic risk stratification of venous thromboembolism in primary care

General practitioners (GPs) play an important role in the diagnosis and management of venous thromboembolic disease (VTE), including deep vein thrombosis (DVT) and pulmonary embolism (PE). With an incidence of approximately 2-3 cases per 1000 persons annually, GPs are encountered with this potentially fatal condition several times each year and with the clinical suspicion even more often. Due to the rather a-specific disease presentation, the identification of patients with VTE can be difficult and GPs are confronted with the dilemma whether or not to refer patients to secondary care for further evaluation. Furthermore, once the diagnosis has been confirmed, the optimal treatment duration (i.e. prognosis) is not always clear-cut, especially in unprovoked VTE.

Diagnostic and prognostic strategies have been developed to aid the risk stratification of patients with (suspected) VTE in the hospital domain. However, these strategies should not be implemented in the primary care domain without evaluation in that specific setting first.

This thesis comprises the evaluation of diagnostic and prognostic risk stratification tools in primary care. First, we quantified the extend of delayed PE diagnosis in primary care in a retrospective chart review. In 26% of confirmed PE cases, diagnostic delay of more than 7 days was present in primary care. Determinants associated with delay were age over 75 years and the absence of typical complaints like pain on inspiration. This subset of patients will not be included in most diagnostic studies on PE due to the initial absence of disease suspicion.

If PE is suspected, the GP has several diagnostic tools at their disposal: prediction models (e.g. Wells rule) or “gestalt” (the GP’s implicit risk estimation), both combined with the fibrin degradation product D-dimer as tested in the patient’s blood. Of all prediction models, we observed the Wells PE rule to be most efficient to rule out PE, and just as safe as “gestalt”. This means that the diagnosis PE can safely be excluded if the predicted probability of PE is low, based on the Wells rule, in combination with a negative point-of-care D-dimer test. Taking into account both costs and health effects, using a point-of-care D-dimer test is thought to be an acceptable alternative to using a lab-based D-dimer test as part of a DVT rule out strategy in primary care.

If VTE is diagnosed, one’s individual prognosis largely determines the optimal anticoagulant treatment duration. In the final chapters of the thesis, the implications of using a formal prognostic prediction model to tailor anticoagulant treatment duration are discussed. We describe the rationale and the design of the Vista trial, a pragmatic multicenter randomized trial on the use of a prognostic prediction model to tailor treatment duration in patients with unprovoked VTE.

In conclusion, the findings of our thesis underline the difficulties in identifying VTE in primary care. Only if the diagnosis is suspected initially, a diagnostic prediction model (preferably the Wells PE rule) can safely be used to identify those patients that require further evaluation in secondary care. The question if patients with unprovoked VTE can benefit from tailored anticoagulant treatment based on a prognostic prediction model will hopefully be answered in the coming years.