WE all understand that clinical suspicion has a low sensitivity and poor specificity for venous thromboembolism.
This does not reduce our distress when a patient dies from pulmonary embolism (PE) after some days or weeks of intermittent or progressing symptoms that, in retrospect, were precursor events.
In days when autopsies were done more often, fatal PE was notoriously unsuspected or misdiagnosed before death. How could we have better recognised these warnings, requested imaging, started therapy and perhaps avoided the fatal outcome?
Coroners are required to make recommendations that “might prevent, or reduce the likelihood of, a recurrence of an event similar to the event that was the subject of the inquest”.
Two cases of fatal pulmonary embolism in South Australia have aroused the interest of the State Coroner.
In two separate incidents during 2012, a 52-year-old man and 63-year-old woman had both attended their community general practice during the weeks before they died, suddenly and unexpectedly.
They were both previously well but had noted calf discomfort attributed to “muscle strain”.
In the man, the diagnosis of deep vein thrombosis (DVT) was considered but thought to be unlikely (a Wells DVT likelihood score of zero or less was quoted in evidence), imaging was not requested, and persisting calf discomfort was managed with physiotherapy for 2 weeks before death.
At autopsy, he had extensive acute and subacute deep leg vein thrombosis plus a massive acute PE.
In the woman, shortness of breath with palpitations and cough was attributed first to stress and then to bronchospasm. DVT was not suspected as a possible cause of calf discomfort.
This autopsy discovered embolism that had recurred during several weeks or months. Leg veins were not examined but DVT was considered the likely source. A month before presenting, the patient had returned by aeroplane from Europe but did not mention travel during consultations. It was also thought she was taking hormone replacement therapy.
The coroner found that early venous imaging for calf pain could have prevented both deaths, as this would have revealed DVT and led to anticoagulant therapy.
The coroner recommended, inter alia, that GPs should, when evaluating calf pain, explore “risk factors including recent long haul air travel” and “symptoms … attributable to pulmonary embolism”, should “not exclude a diagnosis of DVT on clinical grounds alone” or “place undue reliance on the DVT Probability: Wells Score System in attempting to diagnose or exclude DVT or pulmonary embolism” and should have “a low threshold for diagnostic imaging and/or D-Dimer blood testing”.
These recommendations are excellent but do invite commentary.
The prime requirement for diagnosing DVT or PE is clinical suspicion. The common predispositions for venous thromboembolism (VTE) are well known.
We are more likely to suspect DVT or PE during or soon after hospital admission, injury, cancer or long distance travel, or while taking an oestrogen-containing medication. This is natural. But in up to one-third of people with VTE there is no apparent cause, so it follows that we should not rely on predisposition alone to suspect DVT or PE.
The Wells DVT score is an aid for the clinical assessment of patients who attend an accident and emergency, outpatient or community clinic with a first episode of what could be acute DVT (it does not work for hospital inpatients).
Both the score and its clinical use have evolved since it was first published almost 20 years ago. In its most useful form, the score predicts that ultrasound imaging is “unlikely” (≤ 1 points) or “likely” (≥ 2 points) to show DVT; the published prevalences are 5.5% and 28%.
This means the Wells score alone cannot exclude DVT. To do this in outpatients with an “unlikely” Wells score, we must add a D-Dimer test. If the D-Dimer level is normal with an “unlikely” Wells score (the finding in about one in three outpatients with suspected DVT), then we do not need an ultrasound examination, because the risk of a symptomatic VTE during the next 3 months of untreated follow-up is below 1%.
D-Dimer testing is unhelpful when the score is “likely”, and these patients should proceed to imaging, as should those with an “unlikely” score and elevated D-Dimer level.
This diagnostic pathway is supported by evidence-based guidelines and gives us the clinical confidence to minimise low-yield requests for imaging in clinically suspected DVT.
Professor Alex Gallus is professor of haematology at Flinders University School of Medicine and a consultant haematologist at Flinders Medical Centre and the Repatriation General Hospital, Adelaide.