Dec 192014

Put a CT scanner in every U.S. emergency department, add the non-specific signs and symptoms of pulmonary embolism, stir in its potential lethality and morbidity, and line up a few thousand lawyers on the sidelines ready to capitalize on any missed diagnoses, and it's no wonder that the use of CT-angiograms to rule out pulmonary embolism has risen 11-fold between 1998 and 2011.

Most of those extra scans don't seem to achieve much, and may be "overdiagnosis": the mortality rate from PE fell only a trivial amount over the same time period. Meanwhile, doctors may cause a new avoidable cancer with every few thousand CT scans, and pile on millions of dollars in health care costs.

Use of clinical decision tools to rule out pulmonary embolism -- the Wells score, Geneva or modified Geneva score, combined with the use of highly sensitive D-dimer testing, can reduce the use of CT scans. But each testing scheme has its own performance characteristics, meaning at least some false positives and negatives, all of which will vary with the patients (prevalence of PE) studied.

Screening tools are blunt instruments: just being 65, dyspneic and having a heart rate of 75 will get you a "moderate" 20-30% risk for PE on the modified Geneva score. For their part, D-dimer levels increase with age, making false positives common and limiting its utility in the population most likely to present to an emergency room with dyspnea or chest pain.

For all these reasons, using age-adjusted D-dimer values combined with Wells or Geneva scores to rule out pulmonary embolism has been heating up as an area of clinical health services research.

In a BMJ 2013 meta-analysis of 13 cohorts of 12,497 adults > 50 years old at low to moderate risk for pulmonary embolism, using age multiplied by 10 µg/L as the upper limit of normal for D-dimer (instead of the local lab's cutoff) improved the specificity of D-dimer significantly while maintaining a sensitivity of 97% for pulmonary embolism.

However, most studies assessing decision tools to rule out PE in patients with less than high clinical probability have not scanned all participants with CT-angiography, making it impossible to know how many small PEs were missed. (Most have used "clinical follow-up" without obvious PE later as the gold standard -- or had no follow-up at all, in low risk patients.)

That's not enough certainty to have changed the standard of care for emergency medicine physicians in the U.S.

The question is brought into clearer view by a study in the December Chest. Woller et al started with a subset of 934 patients older than 50 (from a larger trial) at low risk for pulmonary embolism, who all got CT-angiograms nonetheless. They were all also followed out to 90 days, with subsequent CT-angiograms for signs/symptoms consistent with PE. Their D-dimer results and outcomes were examined retrospectively:

  • Using the standard cutoff value for D-dimer (non-age-adjusted) would have resulted in no missed diagnoses, at the time of presentation or at 90 days.
  • Using the age-adjusted cutoff resulted in 2 missed diagnoses of pulmonary embolism at the time of presentation (0.7% false negatives who would have been wrongly "ruled out").
  • The age-adjusted D-dimer approach also would have "ruled out" 2 more subjects (another 0.7%) whose CT-angiograms were initially negative, but who developed verified PEs within 90 days.

Those last 2 cases are important -- since in the U.S., their eventual PE diagnoses would lead observers to wrongly consider their initial presentations "missed" diagnoses if CT-angiography had not been done at first presentation (even though they were properly "ruled out" and managed).

The total that would have been viewed as missed diagnoses, then, was 1.5%, but with a confidence range up to 3%. (About 11% of the total cohort had PEs.)

This would have come with the benefit of avoiding about 170 CT scans, or a ~20% reduction.

The majority of the published studies on this question conclude triumphantly but naively that "using our decision-making approach safely reduced the need for CT scans", etc. But there's a reason why few clinicians have abandoned the standard scan-happy approach in favor of these "rule-out" schemes: they all seem to be bound to miss some PEs. This is the first study I'm aware of to suggest just how many.

While an effective 1-3% false-negative rate for pulmonary embolism is low and population health-types might argue it's acceptable, they're not the ones who are blamed for the occasional catastrophic outcome. Expect emergency physicians to continue to order D-dimers and frequently "over-scan" with CT-angiography many patients with positive dimers, until a good prospective randomized trial settles the issue. None are planned on, although this one is interesting.

Read more:

Woller SC et al. Assessment of the Safety and Efficiency of Using an Age-Adjusted D-dimer Threshold to Exclude Suspected Pulmonary Embolism. Chest. 2014 Dec 1;146(6):1444-51.

Diagnostic accuracy of conventional or age adjusted D-dimer cut-off values in older patients with suspected venous thromboembolism: systematic review and meta-analysis. BMJ. 2013 May 3;346:f2492.

"Dealing with D-dimer debacles." Life in the Fast Lane (website).


Get our weekly email update, and explore our library of practice updates and review articles.

PulmCCM is an independent publication not affiliated with or endorsed by any organization, society or journal referenced on the website. (Terms of Use | Privacy Policy)


Age-adjusted D-dimer to rule out PE: coming of age?