Pre-print of the @remap_cap RCT of #anticoagulation for ICU patients w/ #COVID19 adds details but confirms what we learned from the press release:
- no improvement in survival or organ failure w/ therapeutic (TA) vs prophylactic anticoagulation (PA)
- medrxiv.org/content/10.110… 1/
This study is the amalgam of 3 large platform RCTs of TA in COVID19: @remap_cap@ACTIV4a & ATTACC.
Each trial was administered separately but as much as possible they harmonized the design so the results could be analyzed together. (a pragmatic way to enroll more pts faster) 2/
There were some differences:
-REMAP enrolled suspected & confirmed infxn; the others only enrolled confirmed
-choice of anticoagulant varied
-most importantly, the definition of prophylaxis: ~1/2 the sites used standard low-dose heparin, the remainder used “intermediate dose” 3/
The groups were randomized centrally. ACTIV-4a assigned 1:1, the other arms were adaptive so slightly more in PA.
Overall the groups were pretty similar @ baseline: mostly male, median ~60yo, fairly sick (p/f 119)
Most received corticosteroids. Oddly 35-40% were on NIPPV. 4/
The 1° outcome was organ failure support free days (OSFDs) up to day 21. This includes days not on HFNC, NIV, IMV, or ECMO.
I’m not a big fan of lumping these together; being on HFNC is very different than being on ECMO. 🤷♂️
The results are pretty compelling:
-Therapeutic anticoag (TA) didnt increase days free of organ failure support (OSFDs); in fact there were numerically more OSFDs w prophylactic anticoag (PA)
-TA didn’t improve survival
-there was more bleeding w/ TA & more thrombosis w/ PA 6/
Importantly both “standard dose” & “intermediate dose” prophylactic anticoagulation (PA) subgroups trended superior to therapeutic anticoagulation (TA).
Patients not on IMV appeared to do better w/ PA than TA. Big strike against the “you have to anticoagulate early” argument. 7/
They conclude that therapeutic AC is not only likely worse than prophylactic AC, but there is an 81% chance that therapeutic AC actually reduces hospital survival.
The authors (being Bayesians) point out that the small ongoing studies are unlikely to change this conclusion. 8/
Last March as were inundated treating the first wave of people with COVID19, we saw a pattern of thrombosis. It was reasonable to wonder if therapeutic anticoagulation was needed for this new disease.
A year later we’ve studied the problem & we now know the answer: it isn’t.
9/
• • •
Missing some Tweet in this thread? You can try to
force a refresh
Following @remap_cap & #RECOVERY#Tocilizumab results, @NIHCOVIDTxGuide has updated guidelines:
-#Toci + #Dexamethasone now recommended for all ICU pts on IMV, NIPPV, or HFNC
-Toci + Dex recommended for non-ICU pts w/ rapidly increasing O2 needs & elevated inflammatory markers 1/
There are some caveats:
-Toci must be combined with dexamethasone (not given alone; ? harm signal)
-It should be given early (w/i 3 days)
-Toci should NOT be given to people who are already immunosuppressed or who have “an uncontrolled” infxn (e.g. getting worse despite Abx) 3/
🚨Exciting results from the #RECOVERY trial #preprint of #Tocalizumab (Toci) in hospitalized people w/ #COVID19
-reduced 28-day mortality (29 vs 33%; NNT)
-decreased likelihood of requiring MV (33% vs 38%)
-shorter hospital stay (median 20 vs >28 days) medrxiv.org/content/10.110… 1/
They randomized 4116 pts to weight-based Toci vs usual care (UC):
-groups were balanced: mostly male (>65%), older (>60 yo), & w/ comorbidities (>55%)
-most patients (82%) also received dexamethasone
-they received Toci early in hospitalization but were 7-14 days after onset 2/
Notably, only 83% of patients in the Toci group actually received Toci (plus 2.6% randomized to the UC group got Toci); this would decrease the effect size and bias the towards null.
This means their ITT analysis is probably *underestimating* the true effect size somewhat.
It kinda irks me when someone describes a vital sign or lab value as “incompatible with life.”
Here’s a @tweetorial all about the extremes of physiology.
Case #1:
A 10 yo ____ presents with the following vital signs.
T 109F RR 30 HR 300 BP 142/116
Fill in the blank
Answer: 🐓
A chicken's "normal" Temp is 103-110F (w/ HR 220-360) & they live up to 11 yrs.
The Hummingbird would be quite bradycardia (“normal" HR 800-1200 when active)
The Desert ant (Cataglyphis bicolor) has a higher temp (up to 122F!) but doesn't live 10 yrs or have that BP
Case #2:
An *arterial* blood gas is obtained from a ___ showing
pH 7.37 / PCO2 50 / PaO2 20 / HCO3 26
(yup it really is arterial)
First, Abx prescribing was much higher earlier in the pandemic (January 86% vs April 63%) and higher in China (76%) compared to the US (65%) & Europe (63%).
This suggests that overprescribing may be less of an issue currently and in the US.
2/3
Second, only 5 studies (out of 154) reported the Abx duration. We don’t know if Abx was quickly de-escalated (appropriate) vs continued despite (-)cultures (inappropriate).
IMO It’s not wrong to start Abx in sick COVID pts so long as you promptly d/c when cultures are (-)
3/4
🚨BIG NEWS: In January, the unpublished VICTAS trial of vitamin C in #sepsis was stopped after enrolling just 501 of a planned 2000
Now data on clinicaltrials.gov shows why, and it doesn’t look good for #vitaminC. Is this the last🔩in⚰️of the ‘metabolic cure’?
A short🧵
1/
I’ve been hopeful but more than a little skeptical about the 🍹🍋 metabolic cocktail for sepsis (vitamin C + hydrocortisone + thiamine) since the original before/after case series.
I’ve followed this literature closely & have been waiting eagerly for the results of the RCTs.
2/
Thats’s why I was excited to see that VICTAS had posted results. bit.ly/3j3Iatl
The VICTAS trial is the largest (& arguably best) of the vitamin C RCTs: a placebo-controlled, Double-blind RCT done at 43 sites across the US. The 1° endpoint was vasopressor free days.
3/
The authors found that these apparently impressive ANNs were poorly generalizable (i.e., the performance was much worse on a new validation set compared to the training set).
Compare the red vs. green ROC curves. The performance drops from an AUC of 0.99 to 0.7! Yikes! 2/
There’s a reason for this: They used one dataset for all their positive images and a separate dataset for all their negative images.
This is risky for confounding because the model could pick up on any number of differences in CXRs that aren’t clinically meaningful.
3/