The topic of DVT management in ICU patients as only appeared once in the long history of the CICM Fellowship exam. Though Question 20 from the first paper of 2019 asked specifically about upper limb DVT, the focus of what follows is somewhat distracted by the lower limb, as that is where all the action usually is. General principles of management and imaging are generally the same in both scenarios, but there are some peculiarities and differences between the upper and lower body which merit individual discussion. The topic of protecting patients from DVT and VTE is also discussed elsewhere.
The single best reference for this topic would be the chapter by Mal-Dorzi & Arabi from Critical Care Update 2019, which is available through Google Books but is otherwise paywalled by Springer. The excellent review by Kommareddy et al (2002) is freely available, and by itself is enough to answer Question 20 (it's all about upper limb DVT). Another good resource for this article is by Noyer et al (2017), titled "The Arm Is Not The Leg"
Risk factors and incidence of DVT in ICU
It is difficult to say how common DVT is in the ICU, but after reading some of the literature one must conclude that it, at minimum, is much more common (and potentially much much much more common) than in the general hospital community. Hirsch et al (1995), in an ancient but highly cited study, scanned unselected medical ICU patients and a found 33% risk of DVT (48% lower limb, 15% upper limb). Of these patients, only 61% had DVT prophylaxis of some sort. That might suggest that virtuous attention to FASTHUG issues is not entirely protective. The risk of DVT tends to vary quite wildly in more recent studies. For example, Hong et al (2012) found an incidence of 11% among Korean ICU patients, and Miri et al (2017) had a rate of 3.5% in Iran.
It is also difficult to identify exactly which ICU patients are at risk of DVT in the ICU. There's no risk stratification scale which is specific for ICU patients. Sure, one could still apply a conventional DVT risk assessment instrument; and if one does this, one rapidly comes to the conclusion that basically, all ICU patients are in the highest possible risk category for DVT and most will have several high-scoring risk factors. That's what the investigators did in the PROF-ETEV trial (Garcia-Olivarez et al, 2014). They applied the insanely complex, 30-item Caprini model of risk stratification to 777 ICU patients, and calculated that 83% of these patients were fell into the "very high risk" category (8.5-11.6% in-hospital incidence of VTE). It appears that this model, though not ICU-specific, has relatively good validity in the ICU population, insofar as lower scores are associated with a lower risk of VTE, and higher risks are associated with higher risk. Obi et al (2015) confirmed as much when they applied the scoring system retrospectively to a fairly large cohort of ICU patients. The Caprini scoring system is therefore reproduced here, listing risk factors in order of their contribution to the score.
|(1 point)||(2 points)||(3 points)||(5 points)|
- Age <41
|- Age 41-60
- Major surgery
- confined to bed for > 72 hrs
- CVC device
|- Age 61-74
- History of DVT
- Inherited thrombotic disorder
|- Age >75
- Lower limb fracture
-Acute spinal cord injury
Female-specific risk factors:
Risk factors specifically for upper limb DVT
This entity is somewhat less well reported on than lower limb DVT, because it is less common and probably also because of the fact that upper limb thrombi are frequently smaller in mass which makes them less dangerous as emboli. In general the upper limb is held to be an area of the venous circulation less prone to clot formation, perhaps because the arm is a more eloquent limb than the leg. "Even in bedridden patients, the arms are generally less immobilized than the legs, resulting in less venous stasis", is a less poetic comment by Kommareddy et al (2002), but to the same effect. According to Hill & Berry (1990), arm veins have:
- Less stasis
- Lower gravitational stress
- Fewer valves (which promote stasis and act as foci for thrombus formation)
- Higher levels of plasminogen activator and fibrinolytic activity
Unfortunately, there are also a few prothrombotic factors
- They are also more easily compressed, being more superficial
- The total blood flow rate is lower
- They are an attractive site for peripheral and central venous catheters
- They are an attractive site of injections and infusions (which becomes a problem when those are thrombogenic or vesicant, eg. chemotherapy)
- They are often subjected to orthopaedic trauma and are subsequently immobilised in stasis-promoting slings and casts.
The difference in risk profile is discussed in the article by Cote et al (2017). In summary, upper limb DVT risks were different in the following ways:
- Recent surgery was less common in the upper limb DVT patients (12% vs 18%)
- Malignancy was more common (50% had a malignancy, vs 30%)
- Age was less of a risk factor (younger patients developed upper limb DVTs), except where the upper limb clot was provoked by a catheter
- Immobility was much less of a risk factor (19% vs 43% for lower limb DVT)
- Oral contraception or hormone replacement therapy was twice as common among the upper limb DVT group (20% vs 9.2%)
- Of course, upper limb central lines were a major risk factor.
Kommareddy et al (2002) give a broader list of risk factors, which is repeated in the discussion section of Question 20 from the second paper of 2019. It is included because it appears comprehensive, and the source is authoritative (Seminars in Thrombosis and Hemostasis) but the authors do not specify where these data came from.
Risk of catheter-related DVT
Catheter-related DVTs are a specific subset of upper limb DVTs; one might describe these as super-provoked. An excellent UpToDate article is available for the paying customer, specifically concerned with this topic. Because the higher rate for thrombosis is among patients with malignancy, most of the large studies on this topic have been among cancer patients. For example, Saber et al (2011) found the following risk factor associations in their cohort:
- PICC (vs. other catheters, the risk is 5-15%)
- Previous history of DVT
- Subclavian vein insertion site
- Catheter tip not in a greater vessel (i.e. in axillary or subclavian vein)
- Larger catheter size
- An infected catheter
Clinical features of DVT
Clinical features of DVT may be vague, non-specific, and under-reported (the intubated patient will find it difficult to complain of calf pain). It is difficult to diagnose without imaging because of the poor sensitivity and specificity of clinical symptoms and signs. Many of the DVTs in ICU are clinically silent (Williams et al, 2003), in the sense that screening scans often reveal DVTs which could not have been predicted by the intensivist. McKelvie et al (1994) found a ton of clot in the lungs of dead ICU patients during autopsy, most of which were completely unexpected by the clinicians (if the death certificates are anything to go by).
Generic clinical features of DVT:
- Limb swelling
- Cord-like structure on palpation
- Redness due to phlebitis
- Coolness due to poorer perfusion
- Decreased mobility or dexterity of the limb
Features unique to upper limb DVT:
- Neck stiffness
- Facial swelling
- Pemberton's sign (with SVC obstruction)
- Failure to insert upper limb CVCs ("the catheter just won't thread")
There are also a few clinical features unique to catheter-associated DVTs:
- Inability to withdraw blood from the catheter
- Inability to inject into the catheter
- Loss of transducer waveform
Investigations for DVT
- Imaging with ultrasound is noninvasive and requires no radiation. One does not need to visualise the clot as a hyperechoic space-filler: an early clot looks just like blood. Incompressibility of the vessel is the main parameter one looks for.
- Imaging with CT venography is the main approach to diagnosis if ultrasound is not available or for some reason uninformative (eg. the patient is covered in burns dressings). Additionally, CT of the veins is often the only way to map the venous circulation of the mediastinum and the pelvis. A CT venogram would be necessary to explore the venous access of a critically ill patient with upper limb thrombi which are so extensive that their presence makes CVC placement decisions more difficult.
- Imaging with DSA venography is a very eighties (Kinnison, 1986), but may still be done in places where for some reason computed tomography is not available or unreliable (eg. where the scanner is old and the images are of poor quality). Also, apparently it is considered the gold standard for diagnosis of DVT.
Lesser-known techniques include:
- Impedance plethysmography (Hull et al, 1986) where the electrical impedance of the limb is measured with electrodes before and after cuff deflation (the blood, emptying from the thrombosed limb, changes the electrical properties of the limb in a predictably slower way than would occur with nice normal venous drainage).
- Molecular tracer imaging (Houshmand et al, 2014), where the patient is injected with a radioactive tracer which binds to the clot and reveals it on a scintigraphy scan (PET/CT).
Outcomes of DVT
Malato et al (2015) identified the following major issues associated with DVTs in ICU:
- Risk of PE, obviously
- Increased duration of mechanical ventilation
- Increased length of ICU stay
- Increased length of hospital stay
Obviously, pain and poor mobility lead to more analgesia use, falls, and then there's the increased exposure to contrast from all those venograms and the increased risk of bleeding from anticoagulation. In summary, one does not wish to have a DVT in the ICU.
Outcomes of upper limb DVT
The risk of transformation to PE should theoretically be higher with upper limb DVT, given that one tends to flail one's arms around much more than one's legs when one is a delirious ICU patient, and logically that should lead to more frequent clot dislodgement. Contrary to this erudite argument, Cote et al (2017) discovered that patients with upper limb DVT have a far lower rate of PE (9.8% as compared to 25% with lower limb DVT). This factors into the decision to anticoagulate them. Even more interestingly, those patients with upper limb DVT who did have a PE ended up with a higher risk of PE recurrence after therapy, likely reflective of the fact that upper limb DVT has a slightly different risk factor profile (i.e. patients with "unprovoked" upper limb DVT often tend to have nasty prothrombotic conditions such as malignancy).
Management of DVT
Tran et al (2019) had published the most recent guidelines from the Thrombosis and Haemostasis Society of Australia and New Zealand on February 10th of 2019, literally a few weeks before the exam candidates were subjected to Question 20 from the first paper of 2019.
In summary, for proximal and distal DVTs from various causes:
- Distal, provoked by some significant factor, which is now gone:
anticoagulant therapy for 6 weeks.
- Proximal, provoked by some significant factor, which is now gone:
anticoagulant therapy for 3 months.
- Proximal, provoked by some transient non-surgical risk factor:
- anticoagulant therapy for 3-6 months
- Proximal and recurrent:
anticoagulant therapy for "extended period"
- Provoked by malignancy or antiphospholipid syndrome:
anticoagulant therapy for "extended period"
Management of upper limb DVT
Guidelines from the American College of Chest Physicians (Kearon et al, 2012) recommend:
- Isolated brachial vein thrombosis:
no anticoagulation unless symptomatic or associated with a CVC
- Proximal, provoked by some significant factor, which is now gone:
anticoagulant therapy for 3 months.
- Proximal and associated with malignancy:
anticoagulant therapy for 3-6 months.
- Proximal and associated with a central line:
anticoagulant therapy for 3 months if you removed it;
or, anticoagulant therapy until you decide to take it out.
The examiners, in their comments for Question 20 from the first paper of 2019 complained that most of the trainees completely forgot about the role played by CVCs in the development of DVT, and that "almost everyone who remembered wanted to pull the lines out without thought or consideration of the implications." The ACCP agree: there is a need to remember the line, but there is no imperative to remove them:
"In most patients with UEDVT that is associated with a central venous catheter, we suggest that the catheter not be removed if it is functional and there is an ongoing need for the catheter (Grade 2C)"
Aggressive reduction of clot burden
Understandably, in the impatient land of Intensive Care there may be circumstances under which one simply lacks the patience and composure to sit around and wait for the clot to gently resorb over weeks while the patient marinades in heparin. Sometimes the situation calls for urgent action. For these scenarios, there are more violent means.
This was mentioned by the college in their answer to Question 20 as one of the essential points to mention for a passing mark. Though it is well known for the management of PE, the use of thrombolysis for DVT does not enjoy the same level of popularity. That is probably because of the fact that DVT is not a process with immediately life-threatening complications, but the thrombolysis potentially is. However, there are some scenarios when its use becomes more attractive:
- Essential vascular access depends on it (eg. you've run out of central veins and the patient will die without dialysis)
- The clot is massive and precariously placed (eg. in the SVC)
- Long term anticoagulation is not going to be as safe as a short period of coagulopathy while highly supervised (eg. the patient has crippling social issues which preclude the use of long term anticoagulants)
In these scenarios, there are three main options:
- Systemic thrombolysis
- Regional thrombolysis (i.e. you pick a vein distal to the clot and inject a small amount of thrombolytic agent into it)
- Catheter-directed thrombolysis (where a central venous catheter is introduced into the clot from a distant site, and then used to infuse the clot with thrombolytic agent)
What's the evidence to support these cowboy practices, one might rightly ask? Watson et al (2016) were able to scrape together 17 studies which were used to cobble together a Cochrane review with a total of 1103 participants. As one might imagine, the clot was lysed far faster with thrombolysis than with the usual approach, and venous patency was restored much more quickly. This was at a cost of a surprisingly small increase in the risk of bleeding complications (10% vs 8% with standard anticoagulation). The risk of post-thrombotic syndrome (leg swelling, ulcers, etc) was also reduced.
The issue with thrombolysis in DVT is the initiating process, which the thrombolysis does not address. Your vein is made more patent and faster, but you usually will still need to anticoagulate your patient in the short - to - medium term because their antiphospholipid syndrome or malignancy has not gone anywhere. Thrombolysis, therefore, may just add to the cumulative risk of bleeding.
So, when do you thrombolyse them? Vedantham et al (2016) have written a very thoughtful guidance document to help individualise decisionmaking in these scenarios. They recommend the use of thrombolysis in scenarios where:
- The projected risk of bleeding is low
- The risk to life or limb is high (eg. IVC or SVC obstruction)
- The risk of massive PE or post-thrombotic syndrome is great (i.e. iliofemoral DVT less than 14 days of age)
The latter guidance statement does not recommend systemic thrombolysis for DVT. They instead suggest catheter-directed intrathrombus infusion. Distal injection is more likely to lead to useful drug escaping systemically via collaterals. In terms of evidence for catheter-directed thrombolysis, the quagmire of literature has been deep and wide as every medium-large business in the medical instrument industry seems to have created their own brand of catheter with various dispersion-assisting gimmicks (ultrasound, weird vibrating tips, mechanical suckers, etc etc). Fortunately, the ATTRACT trial has shed some light on that (Vedantham et al, 2017). The risk of bleeding was substantially increased, but remained low overall (from 0.3% in controls to 1.7% in the thrombolysis group). The post-thrombotic syndrome severity scores were improved by thrombolysis over the subsequent months, but the patients did not see a big difference in their quality of life. Overall, these data were interpreted as a signal to be cautious, and dampened the enthusiasm for this practice.
IVC and SVC filters
Vena cava filters are discussed in greater detail elsewhere. In short, even though the college mentioned them in their answer to Question 20, one could justifiably omit them from any list of management options for DVT, because they themselves do not manage anything per se. They mitigate the risk of massive VTE by converting it into a risk of IVC or SVC obstruction. The need for anticoagulation is not exactly ameliorated by these devices (in fact the need for anticoagulation increases, if anything). The abundantly demonstrated lack of mortality benefit and the tendency of the filter to migrate randomly and perforate central vascular structures has made this a highly unpopular option. Their use is limited to situations where:
- Anticoagulation is absolutely contraindicated
- PE has occurred in spite of adequate anticoagulation
- There is a large mobile clot distally, and the risk of clot dislodgement is viewed as unacceptable high (eg. the pregnant patient has pelvic vein thrombosis and urgent delivery is planned)
Clot fragmentation and endovascular or open thrombectomy
Obviously, a big vascular surgical procedure is not an attractive option for anybody, and one would rather avoid it. Comerota (2012) wrote about the approaches and outcomes, basically limiting the patient selection to those who have "few alternatives to clear their venous system". Like catheter-directed thrombolysis and all the other aggressive treatments mentioned above, surgical thrombectomy does not obviate the need for anticoagulation.
The advantage of surgical thrombectomy is the possibility of complete and immediate clot clearance as well as stenting or bypassing the affected vessels in a way which rescues the patient from post-thrombosis syndrome. The complications are said to be few: "venous thrombectomy performed by modern era vascular surgeons report little if any associated pulmonary emboli, little if any operative mortality, and a major reduction in the early and late sequelae of iliofemoral venous thrombosis" boasts Comerota. Multiple clever approaches exist, including trying to express the thrombus through a distal venotomy, or giving thrombolytic agents into the ligated vessel and waiting for an hour before releasing the ligature.