Morgenstern, J. What to do about subsegmental pulmonary embolism?, First10EM, July 11, 2022. Available at:
Overdiagnosis is a huge problem in emergency medicine. We know that we are over-diagnosing pulmonary embolism (PE), as the rate of diagnosis has dramatically increased over the decades, but the morbidity and mortality from the disease has remained unchanged. (In other words, although these patients might have a pulmonary embolism, they would have never been harmed by the diagnosis. Instead, we are harming them with unnecessary treatment and an unnecessary diagnostic label.) (Wiener 2011; Bikdeli 2019) Unfortunately, although the problem is obvious, the solution is not. Once a PE has been identified on imaging, treatment is mostly inevitable. One exception might be subsegmental pulmonary emboli. Many have argued that our CT scanners have become too sensitive, and these tiny emboli are probably normal rather than pathologic findings. The most recent American College of Chest Physicians guidelines suggest not treating these subsegmental PEs in patients without a DVT and with low clinical risk for recurrence. (Kearon 2016) However, there is very limited data on the outcomes for patients with subsegmental PEs, which brings us to the SubSegmental Pulmonary Embolism (SSPE) study.
Le Gal G, Kovacs MJ, Bertoletti L, et al. Risk for Recurrent Venous Thromboembolism in Patients With Subsegmental Pulmonary Embolism Managed Without Anticoagulation : A Multicenter Prospective Cohort Study. Ann Intern Med. 2022 Jan;175(1):29-35. doi: 10.7326/M21-2981. Epub 2021 Nov 23. PMID: 34807722
The SubSegmental Pulmonary Embolism study was a prospective cohort study from 18 hospitals in the Netherlands, France, Switzerland, and Canada.
Adult patients diagnosed with an isolated subsegmental PE in the emergency department who were not managed with anticoagulation and who were referred to a thrombosis clinic for management.
Exclusions: Active cancer, history of venous thromboembolism (VTE), oxygen saturation less than 91%, an indication for anticoagulation, more than 48 hours of anticoagulation before enrollment, hospitalization, and pregnancy.
All patients also underwent leg ultrasound on the day of enrollment and at 1 week, and any patient with a proximal DVT was excluded.
None (this is an observational trial).
None (and this will likely be the fundamental limitation when trying to apply these results clinically).
The primary outcome was recurrent venous thromboembolism during the 90 day follow-up period.
They enrolled 292 patients (and don’t seem to explain why they stopped just short of their planned enrollment size of 300). 28 patients had DVTs, of which 20 were anticoagulated and excluded, but 8 had distal DVTs, were not anticoagulated, and were included in the study. Another 6 patients received anticoagulation for other reasons, and were therefore excluded after the fact, leaving them with a total of 266 patients in their primary analysis.
For their primary outcome, there were a total of 8 patients with VTEs in the 90 day follow up period (3.1%, 95% CI 1.6% – 6.1%). 4 were proximal PEs and 4 were proximal DVTs.
2 patients (0.7%) had a major bleeding event, including a fatal case of massive hemoptysis, and 4 patients (1.4%) had minor bleeding events. (Which seems high to me, considering none of these patients were given anti-coagulation.)
The rate of recurrent VTE was lower if you only had a single subsegmental PE (2.1%) and higher if you had multiple (5.7%).
A 3% risk of recurrent VTE is higher than expected with these low risk subsegmental PEs, and may warrant anticoagulation, but there are some problems with this data that limit its clinical application.
They included 8 patients with DVT in this analysis. Although these were distal DVTs, they clearly put the patient at higher risk of finding another VTE within the next 90 days. When we are trying to avoid anticoagulation in patients with subsegmental PEs, I can’t imagine we are trying to do so in patients who also have DVTs, even if they are distal. Only 1 of these patients had a recurrent VTE, but that is equivalent to a rate of 12.5%, which clearly pushed the overall estimate up.
We know VTE rule-in rates are much higher in Europe than they are in Canada. Although the majority of the patients enrolled in this study are Canadian, they don’t include any breakdown of the results by region. It would not surprise me if the VTE recurrence rate was significantly different in Europe and Canada, which would clearly impact the clinical application of these results.
The biggest source of selection bias in this study is the fact that the inclusion criteria specifically look for subsegmental PEs managed without anticoagulation. We don’t know how many patients a physician decided to treat with anticoagulation. On its face, that would seem to bias the recurrence rate down, as the subjectively higher risk patients might have been started on therapy and therefore excluded. However, this is a group that is obviously interested in the topic of subsegmental PE. Therefore, they might have been more aggressive than the average doctor in sending these patients home without therapy, and so the result might be biased towards a higher recurrence rate than we would see in general practice. It is really hard to know.
I am not sure if it matters at all, but it is also worth noting that patients were only included if they were referred to the thrombosis clinic. I am not sure if any emergency doctor is making the decision not to treat these patients unilaterally, without follow up, but if so that would presumably exclude lower risk patients. Similarly, there could be a cohort of patients so low risk that the ED doctor had them follow up with their primary care doctors rather than the thrombosis clinic. In other words, because this cohort only comes from the referral clinic, referral bias could be at play.
There are a mix of biases (verification bias, attention bias, diagnostic suspicion bias) that are likely to combine to inflate the rate of VTE diagnosed in this study. These patients were told they had a PE. They were then “carefully instructed on the signs and symptoms of venous thromboembolism and were instructed to contact study personnel if they occurred.” Furthermore, they were contacted at least 4 times during the 90 day follow-up period to be assessed for symptoms. All of those actions could independently increase the rate of diagnosis. If you are more vigilant about looking for symptoms, you are more likely to find something. I would be interested to see what would happen if you took a group of patients in the general population, and ‘carefully instructed them on the signs and symptoms of venous thromboembolism’. What would the rate of VTE be?
Ultimately, the biggest limitation to this study is the lack of a control group. What is the rate of VTE in a similar population, put through similar screening, and given the same instructions?
More importantly, what are the risks and benefits of anticoagulation in this population? Even if we think that the 3% recurrence rate is real, the risks and benefits will be pretty close. (The maximum possible benefit is 3%, and that is just for recurrence, not for the more important outcomes like death. Treatments are never 100% effective, and the harms of anticoagulation are also of about the same magnitude.) At this point, it is clear that an RCT is warranted to see if anticoagulation is a benefit to these patients.
What does this mean for overdiagnosis?
There is very clear evidence that we are over-diagnosing PE, and harming our patients in the process. (There is also very clear evidence that many of our PE diagnoses are false-positives, which is a related but different issue.) One hypothesis is that subsegmental PEs are more likely to be overdiagnosis than large PEs. The rate of recurrence in this study is high enough to point away from overdiagnosis. I think we clearly need to follow this observational data with proper RCTs to determine the real benefits of treatment in this population.
However, it is quite possible that overdiagnosis occurs just as much with large PEs as it does with the subsegmental ones. In fact, I think this is the more likely hypothesis. A patient’s risk of recurrent VTE is probably much more correlated with their underlying risk factors than it is with the size of a filling defect on a CT scan. Think about the spectrum of patients we assess for PE. A patient with absolutely no risk factors for VTE, who presents with minor pleuritic chest pain and a DDimer just above the threshold probably has an incredibly small chance of recurrence, even if their CT demonstrates a large PE. On the other hand, an obese patient who smokes, has limited mobility, with active cancer, lupus, multiple family members with VTE, and a D-dimer of 10,000 is high risk, even if their CT only shows a subsegmental PE. (This patient would not have been included in this study, but my point is there is a spectrum of risk.) Ultimately, although we like to make decisions based on imaging, it is likely that the risk of recurrent thrombosis is really driven by risk factors.
Unfortunately, that makes it harder for us to address overdiagnosis at the bedside. It would have been easiest if there was a simple dichotomous cut-off, where we could ignore all subsegmental PEs and treat all larger ones. (Medicine is almost never so simple.) Instead, we are left with a much more complex decision. Most subsegmental PEs probably still don’t need treatment, but we may need to consider multiple other factors when making this decision. Conversely, even if most larger PEs do need treatment, it is likely that many of these also represent overdiagnosis, and our treatments are causing more harm than good in some patients.
We clearly still need a lot more research to sort this topic out. For practicing emergency clinicians, I think it is pretty clear that once we have a CT showing a PE, the patient’s course is set. If we really want to limit the harms of overdiagnosis in our own practices, we need to ensure that we are limiting our workups to patients who are truly at high risk of PE, and who will truly benefit from anticoagulation.
At this point, we don’t know the true incidence of recurrent VTE after an isolated subsegmental PE diagnosis. Although there are many sources of bias in this data, the 3% recurrence rate in this study suggests that recurrence is unlikely to be completely negligible. In patients with risk factors, DVTs, and perhaps multiple subsegmental PEs, treatment is probably warranted until we see better data.
In my mind, the best approach is limiting testing to high risk patients in the first place. If I find a subsegmental PE in a patient I thought was high risk for PE, I have no problem treating it. The problem only arises when you really didn’t think the patient had a PE, but got the test for some reason anyway, and found a subsegmental.
Bikdeli B, Wang Y, Jimenez D, Parikh SA, Monreal M, Goldhaber SZ, Krumholz HM. Pulmonary Embolism Hospitalization, Readmission, and Mortality Rates in US Older Adults, 1999-2015. JAMA. 2019 Aug 13;322(6):574-576. doi: 10.1001/jama.2019.8594. PMID: 31408124
Carrier M, Righini M, Le Gal G. Symptomatic subsegmental pulmonary embolism: what is the next step? J Thromb Haemost. 2012 Aug;10(8):1486-90. doi: 10.1111/j.1538-7836.2012.04804.x. PMID: 22672341
Le Gal G, Kovacs MJ, Bertoletti L, Couturaud F, Dennie C, Hirsch AM, Huisman MV, Klok FA, Kraaijpoel N, Mallick R, Pecarskie A, Pena E, Phillips P, Pichon I, Ramsay T, Righini M, Rodger MA, Roy PM, Sanchez O, Schmidt J, Schulman S, Shivakumar S, Trinh-Duc A, Verdet R, Vinsonneau U, Wells P, Wu C, Yeo E, Carrier M; SSPE Investigators. Risk for Recurrent Venous Thromboembolism in Patients With Subsegmental Pulmonary Embolism Managed Without Anticoagulation : A Multicenter Prospective Cohort Study. Ann Intern Med. 2022 Jan;175(1):29-35. doi: 10.7326/M21-2981. Epub 2021 Nov 23. PMID: 34807722
Kearon C, Akl EA, Ornelas J et al. Antithrombotic Therapy for VTE Disease: CHEST Guideline. Chest. 2016. [free full text]
Wiener RS, Schwartz LM, Woloshin S. Time trends in pulmonary embolism in the United States: evidence of overdiagnosis. Arch Intern Med. 2011 May 9;171(9):831-7. doi: 10.1001/archinternmed.2011.178. PMID: 21555660