Overdiagnosis vs. Reasoned Diagnosis: Trust Your Judgment

The Gist:  Here's a paper that confirms that if a clinician doesn't think a patient has a pulmonary embolus (PE), then they probably don't. Basically, another study to make one feel secure in using Wells <2 or clinical gestalt +/- negative d-dimer.  Testing patients who are unlikely to have disease probably does more harm than good (overdiagnosis).  Check out the 7 Step Program for PE workup in June's Annals of EM.


The Study:  Evaluation of Pulmonary Embolism in the Emergency Department and Consistency With a National Quality Measure:  Quantifying the Opportunity for Improvement

• Prospective, multi-site, observational investigating the proportion of avoidable imaging in patients with a low pretest probability (PTP) for PE   
• Reminder:  PTP is the likelihood that a patient has a disease before performing the diagnostic test for that disease.
• Avoidable imaging = imaging in a patient with low PTP, with no D-dimer ordered or negative D-dimer.
• Patients were enrolled if they had an order for a CTPA or V/Q scan by an EM physician during randomly selected "representative shifts."  
• The physician rated the PTP per their gestalt, provided their most likely diagnosis, and documented 74 data points that included the Wells score before receiving the results of the imaging study.  

The Results:

• Primary outcome met:  Imaging deemed potentially avoidable (Wells <2)  in 32% of the patients, 95% CI, 31%-34% (n=1205).
• Sub-analysis showed 29% of patients deemed low risk through clinical gestalt got avoidable imaging
• 38% of low risk patients were imaged and 36% (n=811) of those patients had no d-dimer
• PE found on imaging that was deemed "avoidable" by the above criteria:
• No d-dimer: 1.3% (n=50) 
• Negative d-dimer:  0.2%  (n=8) 

Some Thoughts:

• There's a generally accepted miss rate of 1.8% for PE (1).  The rate of image detected PE in the potentially avoidable images were values were under this range.
• All sites were actively participating in PE research and this may have altered the initial imaging rate (the authors proposed a reduced rate of imaging in this cohort) due to increased awareness of PE related issues.
• The paper doesn't tell us how important this diagnosis is, since it doesn't track a patient-oriented outcome.  Are these clinically significant PEs?  Would treatment do more harm or good in these patients?  This study doesn't answer these more murky, contested questions.
• Authors argue that failure to perform d-dimer testing lead to potentially avoidable imaging.

What's the big deal about testing all of these low PTP patients?

• Dr. David Newman has a great editorial piece that suggests we really consider what it means when we begin to work someone up for a PE, including harms that outweigh the benefits in patients with a low PTP.  Direct and indirect harms from CTPA for PE include:  contrast induced nephropathy, cancer secondary to radiation (1 in 2000 risk of cancer), and major hemorrhage from treatment of the detected PE (6 month bleeding rate of 2.8%).
• The data seems to suggest that pulmonary emboli exist on a spectrum as a disease process.  Thus, the mortality and morbidity of the high clot burden PEs in sick patients are a different story than one that appears in a low risk patient on a super-sensitive CT scan.  The benefits of aggressive therapy are clear in the massive PEs are fairly reasonable in the sub-massive PEs.  This article from the Archives of Int Med is a time trend analysis that seems to suggest that through CTPA we are diagnosing far more PEs but making less impressive advances in preventing mortality and anti-coagulation related complications. Increased diagnosis doesn't necessarily benefit the patient.

So, then, what's a good approach to PE diagnostics?  This editorial in June 2012's Annals of EM proposes the following 7 steps:

1. Accept that you cannot detect every single PE.  Crazy things happen, like a patient having a giant saddle embolus who was low risk by clinical gestalt, PERC negative, and low Wells (Case Report in June Annals of EM), but they're not very common.

2. Recognize that aggressive testing may lead to more patient harm than good.

3. Risk Stratify.  Identify patients with low PTP using your gestalt or PERC.

4. Watchful waiting (or "masterful inactivity").  Advise a 24 hour re-check in the ED or with their PCP. 

5. Document the medical decision making process.

6. Don't worry so much about litigation. This study concludes that physicians ordering CTPA for PE  workup as a defensive medicine practice had less yield in the imaging (likely avoidable imaging).  

7. Spread the word.  Basically, keep the dialogue on testing going.

As a student, I really like this algorithm from EMCrit, since I don't quite have clinical gestalt and I'm new enough to medicine to confirm what I think with algorithms.

On the note of overdiagnosis, LITFL has a recent blog post with some great resources at the end.

References:
1.  Kline Ja, Mitchell AM, Kabrhel C, Richman PB, Courtney DM.  Clinical criteria to prevent unnecessary diagnostic testing in emergency department patients with suspected pulmonary embolism.  J Thromb Haemost. 2004; 2: 1247-1255.
2.  Havig O.  Deep vein thrombosis and pulmonary embolism. An autopsy study with multiple regression analysis of possible risk factors.  Acta Chir Scand Suppl. 1977;478:1-120
3. Green S, Yealy D.  Right-Sizing Testing for Pulmonary Embolism:  Recognizing the Risks of Detecting any Clot.  Annals of Emergency Medicine 2012; 6: 524-526.