Let’s talk perioperative mortality rates and #COVID19.

Huge thanks to the @COVIDSurg Collaborative, publishing the first series of 1128 patients with perioperative #COVID19 infection in @TheLancet.

The take-home number: . %
https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(20)31182-X/fulltext#seccestitle180
(1/)

Almost a quarter of people who get COVID-19 around the time of surgery in this series died. To be frank, that’s awful and scary, whether you’re the person who is about to have surgery or the one who is about to perform it.
(2/)

But the point of studies isn’t to scare us, but to take our blinders off so we can make good decisions. And by the end of this thread, you’ll be able to insert the term “value-of-information” into any conversation and seem 10% more erudite for it.

(3/)

So without further ado, let’s dive into the study. Methods: Open to any center in the world providing surgery. Aim to collect consecutive patients at each institution who were in the Venn diagram overlap btwn those who underwent surgery and those with a #COVID19 diagnosis.
(4/)

Patients were eligible if their #COVID19 diagnosis occurred any time from 7 days before surgery to 30 days after. Surgery was defined as “any procedure done by a surgeon in an operating theatre under general, regional, or local anesthesia”. 1° outcome was 30-day mortality

(5/)

Tbh, when this study was first being designed, one of the big questions was “will we even end up accruing patients?”

Unfortunately, that question turned out to be irrelevant. As the pandemic spread, the number of patients enrolled grew quickly.
(6/)

Between Jan 1 & March 31, 1128 patients at 235 hospitals in 24 countries were enrolled. Most (74%) were emergency surgery patients, a quarter elective. A quarter of patients were diagnosed with #COVID19 before surgery, and the rest in the month following their procedure.
(7/)

About half were surgeries for benign disease, a quarter for cancer, and a fifth for trauma. Three quarters were major surgery. GI/General Surgery and Orthopedics were the most-represented specialties but the procedures performed truly spanned all surgical specialties. (8/)

And now for the headline results. Among 1128 patients, 577 (51.2%) had pulmonary complications, and 268 (23.8%) died. In the paper, mortality rates are provided by specialty, but I’d caution against overinterpretation of these esp those with smaller numbers of patients.
(9/)

Risk factors for increased mortality included male sex, age over 70, ASA 3-5, malignant diagnosis, emergency surgery, and major (vs minor) surgery.

(10/)

What this study doesn’t do – and I would argue at this point that it can’t quite yet – is quantify the excess surgical mortality due to COVID-19. To do this, you’d have to estimate the expected mortality for the 1128 patients undergoing surgery in the absence of COVID.
(11/)

This is a bit complicated but perhaps not impossible. Remember that there are lots of factors that potentially impact perioperative mortality rates.

(12/)

You’d have to account for at least some of these when estimating expected (non-COVID) POMR, but for every factor you account for, you need more patients. Perhaps in future studies this might be possible?

(13/)

We’re left with gestalt, which tells us that these mortality rates are *far* higher than in non-COVID cohorts.

(14/)

Take for example cardiac surgery—34% mortality in this group.
From this randomly selected paper on outcomes following cardiac surgery at an American center, in-hospital mortality was 3.4%.
https://www.annalsthoracicsurgery.org/article/S0003-4975(13)02362-X/pdf
(15/)

So how does this impact patient care? First, it (obviously) impacts surgical decision-making. If you have an active COVID infection, it’s really not the time to get your minimally symptomatic gallbladder out. But that’s an easy one. What about when it gets trickier?
(16/)

Let’s say you have a 6cm thoracic aortic aneurysm that was incidentally found, and let’s say there are high risk features for impending rupture on CT. There’s some COVID circulating in your community but not a ton. Should you undergo surgery? (17/)

Here’s where we get into *value-of-information*. Say you’re the surgeon. How badly do you want to know this patient’s COVID status? Pretty badly, you say. Now, we can actually quantify how badly you want it. Here’s a decision tree.

(17/)

(I haven’t drawn out all the branches, because, microsoft paint, and I don't have TreeAge on my computer)

(18/)

Squares are decisions, circles are chance. You start on the left. If you do a test (and for simplicity, say our test is perfect), you’ll know whether your patient has COVID. You can *then* decide on whether to perform surgery based on what you know about outcomes.

(18/)

Bear in mind that not operating also has a mortality risk, but once you have the COVID information you can weigh the expected outcomes of the operating versus not operating branch and help make the right decision for *your patient*
(19/)

If you decide not to test up front, you make your decision blind. There’s then a chance that you’ll be operating on someone with COVID-19 with resulting high mortality rates.

(20/)

The chance of this happening determines the value of up-front testing. If you’re in New Zealand or Mongolia, the chance of a COVID-19 case being picked up by preoperative screening is small, so the value of this information is somewhat lower to the individual patient.
(21/)

But for the rest of us, the risk is higher. I think there’s an argument to be made for universal preop COVID-19 screening in many places based on this paper. Not only for the bad outcomes that could be averted, but also to minimize the risk of nosocomial spread.
(21/)

Note, with enough information you could actually calculate the numerical value of testing information given various pre-test probabilities of COVID-19, but that’s a degree of sophistication for another day. Suffice it to say, this paper should put us all on guard.
(22/)

Perioperative #COVID19 poses a mortality risk on the order of shock from abdominal sepsis and preoperative testing can provide us with essential information to make the best surgical decision for each patient.
(23/23)