To admit or not to admit:
Does an SDEC service make a difference for venous thromboembolism?

by Ciarán McInerney

Are you more or less likely to be diagnosed with VTE after after being admitted into hospital, or after avoiding an admission because of SDEC?

This was the implied question from a study involving one of our team members. Below we discuss if we now know.

The first few things that you'll need to know to understand the study are:

Venous Thromboembolism: The term "Venous Thromboembolism", or VTE, is when a piece of a blood clot blocks a vein. It is usually treated with a blood thinner to ease the flow of blood and break down blockages.
Thromboprophylaxis: This word is made of two parts. The "thrombo-" bit is the medical term for bloods clots. The "-prophylaxis" bit means prevention because "phylax" means "to guard" and "pro-" means "before" or "in advance". Thus, prophylaxis means to guard ahead of time, i.e. prevent, and so thromboprophylaxis means to prevent blood clots. This is obviously key to treating suspected VTE. As the saying goes: a stitch in time saves nine.

What the researchers found

The researchers looked at records from one hospital Trust in England. They were concerned about venous thromboembolism, VTE, and wanted to know how often people who go to SDEC services will go on to be diagnosed with VTE.

They found that about 1% of patients seen and sent home by an SDEC were diagnoses with VTE soon after. The researchers decided to compare this with the 0.55% of patients who are diagnosed with VTE shortly after being admitted into hospital. This seemed like a intuitive comparison because SDEC services are intended to reduce admissions, generally.

So, the implied question was: Are you more or less likely to be diagnosed with VTE after avoiding an admission via SDEC, or after being admitted into hospital?

What have we learned from this study?

The researchers present two numbers to us: the proportion of people in their hospital who were diagnosed with VTE soon after being admitted hospital, and the proportion after being seen and sent home from an SDEC service. Unfortunately, these two numbers can't tell us much about our likelihood of being diagnosed with VTE after using each service. This is for two types of reasons: statistical and inferential.

Statisticians aren't interested in comparing two numbers to see if one is larger than the other. That is a simple task for anyone who can count. And besides, it will only tell you what is going on with those two numbers, specifically. Statisticians are more interested in comparing large set of numbers (from different but similar sources) because it allows them to study what is going on, generally.

So, while we can easily say that 0.9% is larger than 0.55%, this doesn't translate into saying that your risk of VTE is almost halved by being admitted to hospital rather than being seen and sent home by an SDEC service. The study studied many patients but then made conclusions about one hospital's services. To make conclusions about VTE risk after being seen by different hospitals' services, we would need to study patients from many hospitals.

The second type of reason that these numbers can't be compared is inferential. In other words, there are a few reasons why it is not appropriate to use these data to jump to any conclusion.

For example, consider confounding: Is there a factor that influences a person's likelihood of having VTE and influences doctors' decision to admit the person? If so, then we might be attributing a causal influence to the wrong thing. Are the patients who get admitted fundamentally different to the patients that are seen and sent home by SDEC services? Are a hospital's inpatient services intended for different things than their SDEC services? If so, then we aren't comparing like for like.

Also, there are too many equivocal explanations. In other words, the data can be explained in various ways that lead to different conclusions. For example, we could conclude that being admitted to hospital reduces your risk of developing VTE because 0.55% < 0.9%. But, alternatively, we could say that being seen and sent home by an SDEC service increases the likelihood that we identify your existing VTE because 0.9% > 0.55%. The first explanation makes SDEC services look bad while the second explanation makes SDECs look good.

Where does this leave us?

In short, the study didn't collect enough of the right kind of data to justify "that SDEC patients are at relatively high risk of VTE", as the researchers declare. Their declaration is strongly supported by other information that they acknowledge but did not include in their study design and analysis.

For example, they note that their hospital has a "a strong local VTE prevention program" for patients who are admitted but not for those in SDEC services. Would implementing the prevention program in an SDEC service make the difference disappear? We'd need a particular study to test that. They suggest that people seen by SDEC services have a "high baseline risk due to comorbidities". Would studying subgroups of people with comorbidities make the difference disappear? We'd need a different particular study to test that.

What's next?

In the end, the researchers suggest that it might be worth rolling out routine thromboprophylaxis in SDEC services. Whether or not an SDEC service routinely provides thromboprophylaxis is one of many possible differences between how SDEC services run, and might explain why some work better than others.

This is why, in our project, we are studying the SDEC services in NHS England to understand the ways in which SDEC service differ, and identify which differences make the difference. You can stay informed by signing up to our quarterly bulletin to receive our updates.