VTE Risk and Prophylaxis: Devil in the Details

VTE is a problem – there is definitely a risk of hospitalized patients acquiring and either dying or developing complications from VTE.

The problem is that “hospitalized patients” includes very heterogeneous groups of patients.  We know that the higher risk groups DO benefit from appropriate VTE prophylaxis:

  • Orthopaedic surgical patients
  • General and gynaecological surgery
  • Oncological patients

We also know that some groups are at high risk from complications of VTE prophylactic therapy eg.  GI bleeding, CVA / intracranial bleeds / peripheral vascular dx / skin complications / …

So the question is – how do we decide who needs prophylaxis in the MEDICAL INPATIENTS where the evidence is a lot murkier?  This post is all about the regular medical admissions, the patients who will need a stay on an inpatient ward but are not going to the theatre, have a baby or enter the ICU…  you know, the majority of the patients in most adult hospitals.

This is the most interesting, yet tedious topic in Medicine for hospitalists.  VTE risk assessment and prescribing of prophylaxis is ubiquitous. It is one of those tasks that we do many times a day but rarely think upon too deeply.  It is not an area that I ( nor any I have asked ) have ever received any proper teaching or scientific guidance around.

Every patient that gets admitted gets considered for VTE prophylaxis. Yet, it seems the actual decision-making around our practice is often based entirely on local culture and individual preferences.  I strongly suspect that VTE prophylaxis is very biased by recent events, eg. If a PE is seen in the ED or on the ward, then there will likely be a ripple of aggressive enoxaparin prescribing for a time before it settles back to baseline.

The various guidelines and algorithms employed are based on relatively poor quality and surprisingly small quantity of data.  In fact, there is probably more QI-type data looking into why we do or don’t prescribe VTE prophylaxis, than into the actual efficacy of the intervention!!

Applying the guidelines to an individual patient is tricky. Guidelines try to provide a “one-size-fits-all” approach, where in reality there is a very wide range of risk profiles in the heterogeneous population that is: “medical inpatients”.  Almost all of the published research and protocols suggest that “clinicians should base the decision to prescribe only after considering the individual patient’s risk….” However, none that I have read actually describe how one might go about assessing or assigning a risk profile or enacting the intervention.

There are many factors that go into choosing the most appropriate VTE prophylaxis if any. Few patients fit neatly into a single box.  This is a great example of the clash between protocolized care and common sense.

Ideally, we should all graduate from medical school with the ability to:

  • assess our patient
  • Consider the risks of VTE
  • Consider the contraindications
  • Look for risks of intervention
  • Weigh it all up

The reality is that we tend to follow guidelines or local criteria rather than use our own knowledge to make these small decisions.

I quizzed my trainees about where they “learned” how to prescribe VTE prophylaxis…. and most of them learned it ‘on the job’ by role-modelling and copying the senior doctors… or blindly following local protocols without knowing the theory/ evidence upon which they were based

When I discussed this topic with some international colleagues – it quickly became clear that there is a very different approach and risk tolerance in different countries.  The fact that there are a lot of geographical variations in practice is a hallmark of shaky evidence.

When we look at the numbers – it is clear that we are talking about marginal gains if any for the majority of medical inpatients.  

Anticoagulation is not a pleasant thing to do to somebody – SC injections with bruises and pain are not harmless.  There is some evidence that the only effect of TEDS stockings is to increase the falls rate for old people on the medical wards!

There is also the crucial ability for patients to manage their own risk by mobilising early, doing their exercises etc…  VTE prophylaxis is prime territory for SHARED DECISION MAKING. Patients can be educated and choose to manage their own risk to some extent.

So, what should we do?  How can we tackle this problem?

This is a dive into the data around VTE risk assessment and prophylaxis..  here we go.

To be clear – we are discussing MEDICAL patients, on the ward.  Not Surgical or ICU patients.

So what is the risk of VTE in the average medical inpatient?  Well, that is a tricky question as most of the studies that try to answer this question use routine vascular ultrasound to detect the clots… of course the majority of the patients in these trials are completely asymptomatic and hence may have lived a long and harmless life without ever being aware of or treated for these incidental clots.

When we look at the numbers for actual “clinically detected” detected VTE the number is roughly… 0.34%.  So about 1 in 300 inpatients, taking all MEDICAL admissions. That is pretty low – we ignore much more common risks in our patients (eg. chest pain stratification to this level would definitely lead to more harm than good!)

On the other side of the equation is the potential harms of prophylactic anticoagulation.  Looking at the big data out there  – the best estimate for “major bleeding” is about 0.4% – that is an NNH of 250… basically the same as the average rate of VTE.

Think about this for a minute.  Not all of the clots will be prevented by using prophylactic therapy.  So we can try and work out the NNT (number needed to treat) to balance the risks.  Patients are usually categorised as “LOW” or “High” risk for VTE based on various scoring systems (more on this soon) in the trials into prophylaxis.

The NNT for the LOW-RISK patient is approaching infinite.. there are just not enough clots in this group to make sensible data in any trial I have read.

However, in the HIGH-RISK groups, there is an NNT of 30 to prevent symptomatic DVT and 66 for non-fatal PE. Both of these numbers are lower than the overall NNH of 250.  So, in theory, we should be treating these folk with prophylaxis…  so how do we know who is HIGH-RISK?

Until relatively recently there was really not a lot of solid data to help us sort out the VTE goats and sheep.  As discussed above the decision to treat or refrain was largely left up tot he “spider sense” of the intern.  Fortunately, in recent years, some decent data has been produced to help us on this differentiation of patients into “may benefit” vs. “no benefit, possibly harm” groups.

Firstly, please note that there is a paucity of data to show any significant risk in the first 2 – 3 days.  So you could argue that anyone with an overnight or less than 48-hour stay is by definition low risk.  No study I read even includes these in their cohorts.

A number of clinical decision aids (often called  RAMs ‘risk assessment models’) have been developed and trialled in the last 5 years.  The three  RAMs with the most evidence and best discrimination power are:

  1. Padua
  2. Geneva (don’t confuse with the PE risk score)

This is how the  scores are worked out:

These RAM trials suggest a hazard ratio of between 3 and 6 for most cohorts – that is the risk ratio for developing VTE is 3 – 6 fold in HIGH-risk groups, compared to LOW-risk patients however that was defined.

More importantly, there is actual real-world pragmatic data out there looking at hospitals where these risk assessment tools have been implemented shows that they work… well, what I mean is that they can successfully identify low-risk patients who do not require prophylaxis.

The small STIME trial looked at this and found that despite a reduction in VTE prophylaxis use, there was no change in the rate of VTE observed… that hopefully means that they identified the patients who were never going to benefit and avoided prophylaxis.  We need bigger data on this though as we are still looking for rare events, so big numbers needed to find a true result. 

After looking at the numbers I think that the RAM/decision score with the best discrimination and ease of use is probably the IMPROVE score.  This score also comes with a nice, free online calculator that one can use to balance the risks and potential harms of anticoagulation.  You can find it here:  Uni of Massachusetts , IMPROVE outcomes.

Now I can hear you all complaining: “not another bloody score”.  Whenever we have looked at these scores for other risks in acute medicine they rarely turn out to be better than the Gestalt of an average clinician.  So do we really need to score every single inpatient?

To answer that question I think that you need to audit your current practice around VTE prophylaxis.  The Geneva and Padua validation trials classified between a third and half of patients as “low risk” – so ⅔ or half were HIGH-risk. Note that in these trials the average duration of admission was 8 – 11 days.
It is relatively easy to audit one’s own hospital practice and see if you are in “the ballpark”.

If you want to hear my even simpler system that is not based on hard data, but rather a summation of all the reading I have done then it would look something like this:

Start with reasons to NOT give any prophylaxis:

  1. Is the patient at risk of bleeding acutely? Or are they actively bleeding now?
  2. Are they going to be in the ward for less than 2 days or immobilised in total for < 3 days?
  3. After discussing the relatively small risks & benefits the patient declines therapy

In terms of defining HIGH-risk there are really only a few STRONG risk factors that come up in all the trials:

  1. Previous VTE
  2. Malignancy / myeloproliferative disease
  3. a known thrombophilia / hypercoagulable state

So if they have any of these – they are almost certainly in the HIGH-risk group.  (It should be noted that age (usually > 65 years) is both a weak risk for VTE and for bleeding… so I would ignore “age” as a factor.)

Finally, you are the doctor.  These scores are limited to what the trials included.  Our patients are complex, often there is more information than can ever be presented in a neat table.  So if your patient has a dozen features that make you worry – then you need to weigh that up in your brain.  Just be aware that we are typically biased towards errors or commission over omission.  The urge to treat is strong in our crowd, that is what we do!

If you want to read one single commentary article on this topic then the best I have come across is from Australian physician Dr Alasdair Millar in AMJ, 2014

Let me know how you do VTE prophylaxis.  Do you have a system? Do you have rule of thumb? Help us…



ACCP guidelines VTE prophylaxis – medical pts

Surgical / Orthopaedics are higher risk: see guidelines here BEll et cal
Cochrane’s 2018 review of interventions to increase appropriate VTE prophylaxis
Carpini score in medical patients 2016
“The Caprini risk assessment model was unable to identify a subset of medical patients who benefit from pharmacologic prophylaxis.”
Caprini score in Chinese general hospital inpatients (med and surg)
“There was no incidence difference of venous thromboembolism between surgery and medical patients in the same Caprini level of low (χ(2)=3.58 , P>0.05), moderate(χ(2)=2.89, P>0.05), high(χ(2)=0.46, P>0.05), highest risk(χ(2)=1.61, P>0.05).”

??Padua Prediction Score (Chest 2015)
ENDORSE trial:

Padua Prediction Score vs. Clinical Judgement for VTE

External validation of IMPROVE RAM

Comparison of GENEVA, PADUA, simplified Geneva….

after applying either PADUA or IMPROVE to real hospitals there was no change in rate of VTE… but a large reduction in VTE prophylaxis used

Blood 2016: has the pendulum swung too far?

Missed Doses of Venous Thromboembolism (VTE) Prophylaxis at Community Hospitals: Cause for Alarm Venous Thromboembolism Risk Assessment at a Community Hospital: Adherence to ACCP Guidelines and Cost of Inappropriate Prophylaxis

Association between hospitalization for acute medical illness and VTE risk: A lower efficacy of thromboprophylaxis in elderly patients? Results from the EDITH case-control study.

A Comparison of Enoxaparin with Placebo for the Prevention of Venous Thromboembolism in Acutely Ill Medical Patients (MEDENOX trial )
This cohort represents the highest risk groups we see. 15 – 17% baseline risk in the placebo groups is quite high. Similar to the high risk cohorts in Caprini etc
The risk reduction of 40 mg enoxaparin was about ⅔ – similar to other high risk groups… so yep it makes sense in these patients… how do we identify them?

The LIFENOX Trial LMWH and MOrtality in Acutely Ill Medical Patients….. No difference in death rates…


Fondaparinux for VTE prevention in >60 yo medical inpatients (high risk > 10% baseline VTE rate in placebo). Small trial ~ 320 in each group. Showed about a reduction in death at 32 days of 3 vs 6%,,, that’s a very high rate of death! Once again almost all of the reduction was in asymptomatic VTE during the venography follow up. Randomized, Placebo-Controlled Trial of Dalteparin for the Prevention of Venous Thromboembolism in Acutely Ill Medical Patients (PREVENT Trial)
This one is a mixed bag.. Baseline risk around 4 -5 % is representative of a heterogenous group.
The bottom line showed a reduction of about 50% RR, but almost all of that risk reduction was made up of ASYMPTOMATIC DVT – not actually symptomatic or real life threatening VTE… ask yourself, if they were not in the trial would we notice this “event”…
Philosophically: does an asymptomatic DVT deserve a (ultrasound) sound


A few papers on the natural history / risk of asymptomatic DVTs since it is what we seem to be preventing with prophylaxis….
PREVENT Trial FU – crazy high mortality..

Kahn – good base evidence for Medical Inpatients… can make NNT NNH out of this data

Interesting overview of the history of the guidelines from an Aussie author, Gallus IMJ 2014:

Metanalysis by Dentali 2007 : not enough to justify guidelines:

Metanalysis by Spencer et al 2014:

Editorial by Alisdair MIller from Australia “the Current State of VTE prophylaxis…” excellent risk discussion:

Annals metanalysis from 2011: Quaseem

Speak Your Mind