About a week ago, colchicine seemed to come out of the blue as a possible the treatment of COVID-19 with the COLCORONA trial. For most of us, colchicine wasn’t even on the radar. Now, many are rushing to prescribe it, but does the evidence really support its use?
This blog post will primarily cover the COLCORONA trial, but there was a previous RCT that is worth reviewing first. The GRECCO-19 trial was a multi-center, open-label, randomized clinical trial performed at 16 hospitals in Greece comparing colchicine to usual care in hospitalized adult patients with confirmed COVID-19. (Deftereos 2020) Colchicine was given as a loading dose of 1.5 mg followed by 0.5 mg 1 hour later, and then 0.5 mg twice daily until hospital discharge or 21 days. There were two biochemical primary outcomes and one clinical primary outcome (which is too many ‘primary’ outcomes). Unfortunately, the trial was stopped early due to low enrollment. Furthermore, almost all patients received treatment that would not currently be considered standard of care (hydroxychloroquine and azithromycin). The clinical primary outcome – clinical deterioration by 2 points on an ordinal scale – occurred in 7 patients (14.0%) in the control group and 1 patient (1.8%) in the colchicine group (p=0.02). There were not any statistically or clinically significant differences in the two biochemical primary outcomes (based on troponin and CRP values). Although the clinical outcomes in this trial look promising, conclusions are limited by its open label design, small size, and use of multiple primary outcomes. (Deftereos 2020)
Considering the therapies that we have jumped on this year with far weaker evidence, I am somewhat surprise that colchicine was not already in widespread use. I anticipate that might be changing, considering the somewhat sensational press releases that accompanied COLCORONA. So what does the COLCORONA trial actually tell us?
COLCORONA: Tardif J-C, Bouabdallaoui N, L’Allier PL, et al. Efficacy of Colchicine in Non-Hospitalized Patients with COVID-19. Infectious Diseases (except HIV/AIDS); 2021 Available from: http://medrxiv.org/lookup/doi/10.1101/2021.01.26.21250494
COLCORONA is a randomized, double-blind, placebo-controlled trial.
They targeted what they considered high risk patients with COVID-19 in the outpatient setting. Patients were over the age of 40, not under immediate consideration for hospitalization, and with at least one high risk criteria:
- Age over 70
- Hypertension (systolic BP 150 or above)
- Known respiratory disease
- Known heart failure
- Known coronary disease
- Fever of at least 38.4 Celsius in the last 24 hours (not sure how this is relevant)
- Dyspnea at the time of presentation
- Bicytopenia, pancytopenia, or the combination of high neutrophil and low lymphocyte counts
Patients needed to have COVID-19 confirmed by PCR testing, although there were some patients enrolled early in the trial without confirmatory testing because testing wasn’t available.
There are some exclusion you can look at if you decide to start prescribing colchicine, but I don’t think there is a big role for colchicine, so I won’t list them here.
Colchicine (0.5 mg orally, twice daily for 3 days, then daily for 27 days).
The primary outcome was a composite of death or hospitalization due to COVID-19 infection in the 30 days following randomization.
They included 4488 patients, of the 6000 they had planned to enroll. Patients were a mean age of 55, about half were women, and they were enrolled about 5 days after symptom onset. The groups look balanced based on the data they present.
The primary endpoint was negative, occurring in 4.7% of the colchicine group and 5.8% of the placebo group (OR 0.79; 95.1% CI 0.61 to 1.03; P=0.08).
Individually, neither mortality (0.2% vs 0.4%; OR 0.56; 95% CI 0.19 to 1.67) nor hospitalization due to COVID-19 (4.5% vs 5.7%; OR 0.79; 95% CI 0.60 to 1.03) were statistically different. The need for mechanical ventilation was also not statistically different between the groups (OR 0.53; 95% CI 0.25 to 1.09).
So where does all the excitement from from? They report some statistically significant outcomes in the subset of 4159 patients with COVID-19 confirmed by PCR, but such hypothesis generating secondary analyses are probably unhelpful to us, as the pandemic will be over before anyone is able to test them. (Mortality still wasn’t statistically significant, but “hospitalization due to COVID-19” was.)
Serious adverse events were somewhat lower with colchicine, but I don’t see this term defined anywhere in the currently available manuscript.
Overall adverse events were much higher in the colchicine group (24% vs 15%), driven largely by diarrhea.
COLCORONA represents yet another example of bad science through press release during the COVID-19 pandemic. The press release made dramatic sounding claims by only speaking in relative risks. The press release leads with a secondary analysis, with no mention of the fact that the primary outcome of the trial was negative. The press release uses the phrase “approached statistical significance” as a bad euphemism for “the primary outcome was not statistically significant”.
Despite some pretty dramatic claims being made in the media, this is a pretty un-exciting trial. The primary outcome was negative. Even if you believe the secondary analysis, the benefits they are claiming are tiny. Of course, the severity of this pandemic would allow me to embrace therapies with NNTs over 100, but considering the significant limitations of this data, benefit is far from certain.
One important caveat: we still only have a pre-print version available. The manuscript has not been peer reviewed. That is becoming commonplace during COVID-19, but never before would we change clinical practice based on research that has not been peer reviewed. Be cautious.
Perhaps the biggest issue with the trial is that it was stopped early, and they didn’t stop based on any pre-specified stopping point. There are some logistical issues mentioned in the manuscript, but mostly it sounds like the researchers felt they ‘needed to provide healthcare systems with these results as soon as possible’. Stopping trials early always increases the risk of bias, and the reason for stopping here suggests a degree of optimism that also suggests a high risk of bias. Even worse, they stopped the trial while it was still negative, and now we are left with the possibility that the study was simply under-powered to detect a real difference, and given the timing of the pandemic, we will probably never get an answer.
The primary outcome is also problematic. Composite outcomes are always an issue, in that they can combine important outcomes with much less important outcomes. However, my bigger concern is the use of “hospitalization due to COVID-19” instead of just “hospitalization”. This is very similar to the incredibly problematic practice of using disease specific mortality instead of all cause mortality. It is a nonsensical outcome that automatically biases the results in favour of the intervention. It excludes any hospitalizations that occurred as a result of adverse events, and introduces significant subjectivity into an otherwise objective outcome. It just doesn’t make sense.
The overall rate of hospitalization is not listed anywhere in this manuscript. It is entirely possible that patients taking colchicine were actually admitted to hospital far more often than those taking placebo, but that their admission diagnosis was simply something other than COVID-19 (such as diarrhea, acute renal failure, rhabdomyolysis, or one of the many other side effects of colchicine). It is impossible to know based on this manuscript.
Thus, I think you need to completely ignore the outcome “hospitalization due to COVID-19”. That leaves us just with mortality (which was nowhere close to statistically significant, and even if it was, only had an absolute difference of 0.2%, or an NNT of 500) and need for mechanical ventilation (also not statistically significant, and with an absolute difference of 0.5%). This is not a game changing study.
There are other colchicine trials underway, but none as large as COLCORONA. Of note, there are two trials listed as completed on clinicatrials.gov (here and here), but which haven’t been published yet, which suggests they were negative. Publication bias is an important consideration.
There was a previous, low quality RCT that looked somewhat promising. That might have increased your pretest probability that colchicine could help. But if we had seen 2 subsequent negative RCTs, we would probably be pretty negative about colchicine at this point. That is why publication bias matters so much. It shapes the overall literature and affects our interpretations of new trials. For COLCORONA, it probably doesn’t matter too much. Despite the incredible publicity, this is clearly a negative trial. It is incredibly unlikely that colchicine is actually saving lives. It is unfortunate that this trial was stopped early, because we clearly need more research, and now by the time it can be completed, it will likely be irrelevant.
Although it is possible that colchicine has a tiny benefit among outpatients with COVID-19, this is a negative trial. The benefit claimed is based on a secondary analysis using a subjective and biased outcome. Based on the available research, it is unclear if there are any real benefits from colchicine, or whether benefits might outweigh risks, and so colchicine should not be prescribed routinely for COVID-19.
Deftereos SG, Giannopoulos G, Vrachatis DA, et al. Effect of Colchicine vs Standard Care on Cardiac and Inflammatory Biomarkers and Clinical Outcomes in Patients Hospitalized With Coronavirus Disease 2019 JAMA Netw Open. 2020; 3(6):e2013136-. https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2767593
COLCORONA: Tardif J-C, Bouabdallaoui N, L’Allier PL, et al. Efficacy of Colchicine in Non-Hospitalized Patients with COVID-19 [Internet]. Infectious Diseases (except HIV/AIDS); 2021 [cited 2021 Jan 29]. Available from: http://medrxiv.org/lookup/doi/10.1101/2021.01.26.21250494