Last week, I discussed chloroquine (and hydroxychloroquine) for COVID, and the complete lack of convincing evidence. There are various other treatments that have been proposed, but none of them have strong evidence. However, seeing as COVID is a virus, I guess we should take some time to look at antiviral medications. Based on their name, it at least sounds like they should work. (They probably don’t. The best scientific answer is still that the treatment for COVID is high quality supportive care.)
Lopinavir/ritonavir, which goes by the trade name Kaletra, is a combination therapy typically used to treat HIV. It has been studied in the lab, and has in vitro effects against SARS and MERS, and also seems to be effective in an animal model. There were a couple retrospective looks at this combination in SARS which seemed promising, but the drug was generally used in combination with ribavirin so the results are messy, and the retrospective nature of those trials is a significant limitation. (Chu 2004; Chan 2003) There are also numerous case reports of lopinavir/ritonavir being used in a small handful of COVID-19 patients, but without controls I can’t find any clinical value in these reports. (Young 2020, Kim 2020, Lim 2020)
Until this month, there was no reliable clinical data, although there is an RCT underway in MERS patients (NCT02845843). Despite being a novel infection, the sheer numbers we have seen with COVID-19 means that they were able to design, run, write-up, and publish an RCT in just a couple months, while the MERS RCT has been underway for almost 4 years. Of all the COVID antiviral therapies, this is the only one with a high quality trial, so we might as well start here:
Cao B, Wang Y, Wen D, et al. A Trial of Lopinavir-Ritonavir in Adults Hospitalized with Severe Covid-19 [published online ahead of print, 2020 Mar 18]. N Engl J Med. 2020;10.1056/NEJMoa2001282. PMID: 32187464 [open access article]
Methods: This is a single-center open-label RCT from Hubei, China. (They did not create placebos because of the speed with which the trial was put together.) They included adult patients with a positive PCR for COVID-19, pneumonia on chest x ray, and an oxygen saturation of 94% or less on room air (or a PaO2:FiO2 <300). Physicians were allowed to exclude patients from the trial if they thought “it was not in the patient’s best interest”, which introduces possible selection bias, but they did use an appropriate allocation concealment procedure. They compared lopinavir-ritonavir (400 mg and 100 mg by mouth twice daily for two weeks) to standard care. They assessed patients using a 7 category scale:
- Not hospitalized, with resumption of normal activities
- Not hospitalized, but unable to resume normal activities
- Hospitalized, not requiring oxygen
- Hospitalized, requiring oxygen
- Hospitalized, requiring high-flow nasal oxygen, noninvasive ventilation, or both
- Hospitalized, requiring ECMO, mechanical ventilation, or both
Results: They included 199 patients, of whom 95% actually received the assigned treatment.The median age was 58, 60% were males, and the median time interval between symptom onset and randomization was 13 days. This was a very sick group, with a 22% mortality rate overall. For their primary outcome, the time until the patient improved by 2 points on that scale or were discharged from hospital, there was no difference at all. The median time to improvement was 16 days in both groups. There was also no difference in time to deterioration (HR 1.01; 95% CI 0.76-1.34). There was no difference in mortality (19.2% vs 25.0%, ARR -5.8%, 95% CI -17.3 to 5.7%). Although this is not statistically significant, a lot of people have looked at the absolute difference here and become excited. I think there may be some leeway in what stats we should require in a pandemic (if we required a 5,000 person trial, it might not be completed until the pandemic was over). However, there are multiple methodologic problems with this trial, so I think it is difficult to put too much faith in this number. Furthermore, the stats say that this finding is consistent with lopinavir-ritonavir increasing mortality by 5.7%, which should really limit the excitement for this medication. There was also a borderline statistical decrease in ICU length of stay (6 vs 11 days, ARR -5 days, 95% CI -9 to 0 days). However, it isn’t clear to me how that is possible, when the primary outcome includes the things that would have kept these patients in the ICU and was negative. More importantly, the difference in length of ICU stay was only found among the patients who died, so it isn’t actually a very good surrogate for anything. As an interesting note, virus RNA was still detected in 41% of the patients at the end of the trial (28 days), and they didn’t find any significant antiviral effect from the antiviral medication.
Comments: There are a number of issues with this trial. It was unblinded, and the primary outcome is only quasi-objective. They adjusted their sample size on the fly. Now, it is very difficult to decide on a sample size when very little is known about a disease, so we need to give them some leeway, but they did change their sample size partway through, and then decided to stop the study not because they reached a new sample size, but because there was a new agent available to test (remdesivir) that they were more excited about. The biggest concern here may be the timing of therapy. This will be an issue for all antivirals, and so I will come back to it at the end. Patients weren’t treated until 13 days into their course. That may be too late to show a benefit. At that point, viral replication is not the major physiologic issue (we believe patients are experiencing cytokine storm at this point), so it isn’t clear that we should expect an antiviral to work. However, if we give the antivirals earlier, we will have to treat a lot of patients who were never going to get severely ill, which will significantly increase the number of patients harmed without adding benefit.
There is evidence that remdesivir works against MERS in vitro and in animal models. (Sheahan 2020) However, as we saw with chloroquine and lopinavir/ritonavir, in vitro activity doesn’t necessarily translate into clinical benefit. That is it. There is no available clinical data that remdesivir helps patients with COVID-19. There are multiple RCTs underway, and I will update you when they become available, but for now there is no evidence of benefit, and remdesivir should not be used outside of one of those clinical trials.
UPDATE: There was an observational trial published in the New England Journal of Medicine (always a red flag that you might be looking at advertising rather than science) looking at remdesivir for COVID-19. (Grein 2020) Because of the lack of a control group (along with selection bias, a lack of a primary outcome, and extensive financial conflicts of interest), I don’t think this study provides us with any scientific value at all. If you want to read more about it, check out the post “Eleven reasons the NEJM paper on remdesivir reveals nothing” on PulmCrit.
UPDATE (April 30): We have now seen the first published RCT of remdesivir. (Wang 2020) It is a double blind, placebo controlled trial. They included 236 adult patients admitted with confirmed COVID-19 pneumonia at 1 of 10 hospitals in Wuhan, and randomized them to remdesivir (200 mg on day one and then 100 mg for 9 days) or placebo. Their primary outcome was time to improvement by 2 points on a 6 point scale or discharge from hospital alive. There was no difference (21 vs 23 days; HR 1.23; 95% CI 0.87 – 1.75). There was also no difference in mortality. The trial was stopped early, which may have resulted in it being under-powered, but it is clear that remdesivir is no wonder drug. You can read more here.
It is worth knowing that there is a second RCT that has been completed – the ACTT trial. We know about it because of press releases, but the study has not actually be published yet, so it is hard to say anything about the results. I will update when that trial is actually published.
UPDATE (May 27): The ACTT trial is now published. The results look promising, although there are some issues with the trial, such as a change in primary outcome and a decision to stop the trial early. Patients taking remdesivir had a shorter time to recovery (11 vs 15 days). I think that result is very promising, but given the issues with the trial, and my low pretest probability, I still think it is more likely than not that remdesivir doesn’t help. You can read the full critical appraisal here.
Ribavirin and Interferon
Interferons are a cytokines with nonspecific antiviral properties. There are 12 in vitro studies of interferon in SARS, and all reported antiviral activity. (Stockman 2006) Ribavirin is a guanosine analog with antiviral activity against multiple RNA viruses. There are 6 in vitro trials of ibavirin against SARS, and 4 of those trials found an antiviral effect. (Stockman 2006) Of 24 clinical reports of ribavirin use in SARS, 20 were considered inconclusive, and 4 demonstrated harm. Reports of hemolytic anemia with ribavirin use in SARS varied between 36 and 61%. (Stockman 2006) There were 2 trials of interferon with SARS patients, and neither demonstrated a benefit. (Stockman 2006) There is a retrospective cohort of 144 MERS patients treated with the combination of ribavirin and interferon, and mortality was higher than it was in controls (74% vs 62%, p=0.02). (Arabi 2019) It also wasn’t associated with any faster viral clearance. Despite that, there are a couple trials underway for COVID-19 (NCT04254874).
There is a guidance document from British Columbia. For most of these treatments, it recommends against use of the medication clinically, outside of an RCT. The guidance for ribavirin and interferon is much stricter: “ Strongly recommend against use of ribavirin and/or interferon for risk of harm.”
Even though oseltamivir probably doesn’t even work against influenza, and there is evidence from SARS that it does not even have in vitro activity against coronaviruses, it has been used in COVID patients. (Tan 2004) In a single-center chart review of 99 patients from Wuhan with COVID-19, 75% received some kind of antiviral therapy, the most common of which was oseltamivir. (Chen 2020) There are trials underway (NCT04255017), but there is no reason to prescribe oseltamivir at this time.
Balozavir is another antiviral agent that has been tested for influenza. (I won’t get into the question of whether it provides any benefit for influenza in this post.) It has been suggested that it might work against COVID-19, but there is no clinical data to back that claim up. There are 2 ongoing Chinese RCTs (ChiCTR2000029548 and ChiCTR2000029544). It should not be used clinically, outside of an RCT, at this point.
Various other medications
There are a number of other antivirals that are being tested. At this point (March 28), all are supported by worse evidence than those already discussed. As far as I can tell, none of the medications below even have in vitro data or basic science supporting their use against COVID-19.
- Darunavir/cobicistat (Prezcobix) is an HIV antiviral. There are at least 2 COVID-19 RCTs underway. (NCT04252274; ChiCTR2000029541)
- Azvudine is primarily used against hepatitis C, but has also been tested in HIV. There are at least 4 COVID-19 RCTs underway. (ChiCTR2000029853).
- Favipiravir is an experimental antiviral that has been considered for influenza, West Nile, and Yellow fever, among many other viral illnesses. There are at least 4 COVID-19 RCTs underway.
- Triazavirin is a potential influenza antiviral. There is at least 1 COVID-19 RCT underway.
- Umifenovir is a medication that apparently is widely used for influenza in Russia. (It may even be available over the counter there?) There is not a lot of English language data, but Wikipedia says that “the Russian Academy of Medical Sciences stated that the effects of Arbidol (umifenovir) are not scientifically proven.” There are at least 5 COVID-19 RCTs underway.
- ASC09 doesn’t even have a name yet. There are at least 4 COVID-19 RCTs underway.
- Danoprevir is a protease inhibitor used against hepatitis C. There is at least 1 COVID-19 RCT underway.
- There are also RCTs underway looking at anti-parasitic medications such as suramin, and antibiotics such as carriomycin, and of course azithromycin, in addition to many other trials looking at various anti-inflammatory medications, vitamins, and any form of pseudoscientific woo you can think of.
It is good to see vigorous scientific activity in the setting of a pandemic, but we need to make sure our scientific standards are maintained. Although it is difficult to create trials with our usually expected rigour, doing anything else will leave us with faulty and unreliable answers. In particular, looking through this literature, I am concerned about the sheer number of RCTs underway without clear biologic plausibility. We are skipping a lot of our usual laboratory work. We seem to be willing to test anything we have available. That creates a problem when trying to assess the pretest probability for many of these medications. There are so many trials underway that some are guaranteed to have a p value less than 0.05 by chance alone. The mindset of the medical community during this pandemic means we are more likely to run with any positive results, even though a survey of the vast number of trials underway should warn us that positive trials are more likely to represent false positives than true benefit for our patients. I worry that we are setting ourselves up to be wrong.
Timing of antiviral therapy
When it comes to antiviral medications, the biggest question in my mind is when they should be given. We try to start antivirals for shingles as soon as possible. People rush to treat influenza with oseltamivir (although likely without net benefit). The reason that we rush is that the virus is replicating rapidly early in the infection. If you wait until the patient is admitted to hospital, the pathophysiology has changed. Clinical symptoms are based more on inflammatory effects than viral replication. Therefore, giving antivirals at this stage might simply be too late, which could explain the negative trials thus far. However, as of right now, we don’t have any method to reliably predict which patients will require ICU management. Therefore, if we wanted to give antivirals earlier, we would have to give them to a large number of patients who would never benefit, but would be subjected to the harms. It isn’t clear whether we will be able to overcome this timing problem to create a net benefit for our patients.
Harms of prescribing
A common refrain in medicine is “what’s the harm?” Why not prescribe experimental medications? Why not give patients a shot? The problem is, every medication causes harm, and those harms are likely to be amplified in sick patients. (A case of anaphylaxis in the community may be relatively minor, but anaphylaxis in an ICU patient already in septic shock could be the factor that pushes the patient towards death.) Furthermore, we are already hearing about critical shortages of these medications. Patients who rely on them for proven indications are being robbed by our overzealous use of them in COVID-19, with no evidence at all that they are helping. I think “what’s the harm” is a bad argument.
At this point, I don’t think any of these antiviral medications should be routinely used for COVID-19 patients. Many have been shown to have harms, and none have been demonstrated to help our patients. I think it is reasonable to enroll patients in RCTs, but when those RCTs are published, we will need to interpret them within the context which they were created: hundreds of lower quality studies of agents with very limited biologic plausibility. Some will be positive by chance alone, but that doesn’t mean we are truly helping patients.
On this topic, as I was finishing this post, we received a message from the College of Physicians and Surgeons of Ontario. They state: “There is no evidence that anti-malarial, antibiotic and HIV and other anti-viral medications are effective in prophylaxis or treatment of COVID-19, yet the prescribing of these drugs has skyrocketed, leading to shortages for patients who need these medications for other conditions. We ask that you not prescribe these drugs to treat COVID-19 pending further evidence-based recommendations.”
REBEL EM: Lopinavir-Ritonavir in Hospitalized Patients with Severe COVID-19
PulmCrit- Is Lopinavir/Ritonavir down and out?
PulmCrit – Preview of Lopinavir/Ritonavir efficacy in COVID-19?
Useful table summarizing COVID evidence from the American Society of Health-System Pharmacists
Some general information about what is being studied from the CEBM
Arabi YM, Shalhoub S, Mandourah Y, et al. Ribavirin and Interferon Therapy for Critically Ill Patients With Middle East Respiratory Syndrome: A Multicenter Observational Study . 2019;
Cao B, Wang Y, Wen D, et al. A Trial of Lopinavir-Ritonavir in Adults Hospitalized with Severe Covid-19 [published online ahead of print, 2020 Mar 18]. N Engl J Med. 2020;10.1056/NEJMoa2001282.
Chan KS, Lai ST, Chu CM, et al. Treatment of severe acute respiratory syndrome with lopinavir/ritonavir: a multicentre retrospective matched cohort study. Hong Kong medical journal = Xianggang yi xue za zhi. 2003; 9(6):399-406. [pubmed]
Chen N, Zhou M, Dong X, et al. Epidemiological and clinical characteristics of 99 cases of 2019 novel coronavirus pneumonia in Wuhan, China: a descriptive study The Lancet. 2020; 395(10223):507-513.
Chu CM. Role of lopinavir/ritonavir in the treatment of SARS: initial virological and clinical findings Thorax. 2004; 59(3):252-256.
Grein J, Ohmagari N, Shin D, et al. Compassionate Use of Remdesivir for Patients with Severe Covid-19 [published online ahead of print, 2020 Apr 10]. N Engl J Med. 2020;NEJMoa2007016. doi:10.1056/NEJMoa2007016 PMID: 32275812
Kim JY, Choe PG, Oh Y, et al. The First Case of 2019 Novel Coronavirus Pneumonia Imported into Korea from Wuhan, China: Implication for Infection Prevention and Control Measures J Korean Med Sci. 2020; 35(5).
Lim J, Jeon S, Shin H, et al. Case of the Index Patient Who Caused Tertiary Transmission of Coronavirus Disease 2019 in Korea: the Application of Lopinavir/Ritonavir for the Treatment of COVID-19 Pneumonia Monitored by Quantitative RT-PCR J Korean Med Sci. 2020; 35(6).
Sheahan TP, Sims AC, Leist SR, et al. Comparative therapeutic efficacy of remdesivir and combination lopinavir, ritonavir, and interferon beta against MERS-CoV Nat Commun. 2020; 11(1).
Stockman LJ, Bellamy R, Garner P. SARS: Systematic Review of Treatment Effects PLoS Med. 2006; 3(9):e343-.
Tan EL, Ooi EE, Lin C, et al. Inhibition of SARS Coronavirus Infection In Vitro with Clinically Approved Antiviral Drugs Emerg. Infect. Dis.. 2004; 10(4):581-586.
Young BE, Ong SWX, Kalimuddin S, et al. Epidemiologic Features and Clinical Course of Patients Infected With SARS-CoV-2 in Singapore JAMA. 2020;
Wang Y, Zhang D, Du G, et al. Remdesivir in adults with severe COVID-19: a randomised, double-blind, placebo-controlled, multicentre trial The Lancet. 2020; [article]
Morgenstern, J. Antivirals for COVID19 – Not ready for clinical use, First10EM, March 30, 2020. Available at:
4 thoughts on “Antivirals for COVID19 – Not ready for clinical use”
I think it’s a bit disingenuous to say that we should take the sheen of lopinavir/ritonavir because the 95% CI includes worsening mortality by 5.7%, when it also includes improving mortality by 17%.
Also, you shouldn’t say there the trial demonstrated that there was no difference in mortality, that is an incorrect interpretation of the statistics. The trial failed to demonstrate that the difference in mortality was “statistically significant” (whatever that means). It did NOT conclude the mortality difference was zero.
Furthermore, the study was not powered to detect differences in secondary outcomes, so it’s not like they were even trying to rule this outcome out.
All this is not to say that we should start handing out lopinavir/ritonavir like candy to these patients. Just that it should definitely not be ruled out based on one very flawed trial.
Thanks for the comment.
I only include discussion of 1 side of the 95% confidence interval because I think that is the important part of the conversation. People are rushing to prescribe this drug based on a negative study with an underpowered, but statistically negative secondary outcome. The point of the discussion is that people who are excited about the point estimate of 5% improvement need to know that it is actually statistically equivalent to 5% worse.
Thanks for the article and the discussion. I agree with you that I don’t think people should be rushing to use this medication. I am also worried about the opposite behavior, which I am also seeing: people burying this medication, saying “it doesn’t work.” I think that’s also dangerous. And I think comes in part because of a misinterpretations about statistics / statistical language.
The term “statistically negative” might imply that an outcome was statistically shown to be “negative.” That is not what we have for this secondary outcome. Statistically, this trial was inadequately powered to determine the statistical nature of this outcome. Statistically inadequate or ambiguous, sure. But statistically negative? I think that’s wrong.
And then “statistically equivalent” is also misleading. The 95% confidence interval gives a range in which you believe the true outcome to be. It does NOT imply or tell you that all values are equally likely within the range. And in most statistical models / distributions we see, the values at the end of these intervals are the least likely to be true. The measured value of the outcome is possible, as are all other values within the 95% confidence interval. But that does NOT mean they are all “statistically equivalent.”