Azithromycin monotherapy shows no benefit for COVID-19: RECOVERY study reports 1493 further deaths

Simon Collins, HIV i-Base

On 14 December 2020, the large randomised UK RECOVERY study announced that azithromycin had no benefit for the treatment of COVID-19. [1]

Previous results from this study included positive results for dexamethasone in last-stage infection but negative results for hydroxychloroquine and lopinavir/r.

Azithromycin had been chosen as a macrolide antibiotic that in additional to antibacterial activity can also reduce proinflammatory cytokines and has in vitro antiviral activity, including against SARS-CoV-2. Macrolide antibiotics are used to treat bacterial pneumonia and chronic inflammatory lung disease.

On 27 November 2020, the trial steering committee closed the azithromycin arm based on ‘sufficient patients having been enrolled to establish clearly whether or not the drug had a meaningful benefit’. This shows that the timing was based on predetermined statistical calculations, rather than close observations from the Data and Safety Monitoring Board (DSMB).

Between 7 April and 27 November, 7,764 participants were randomised 1:2 to either azithromycin (n=2582) or standard of care (n=5182). However, more than 16,000 people were recruited to the RECOVERY study and just over 7000 were excluded from this randomisation because of contraindications to azithromycin or unavailability of drug. Roughly 15% of both the azithromycin and control participants were also enrolled in a second randomisation to another experimental drug.

Azithromycin (500 mg) was to given by mouth, nasogastric tube, or intravenous injection once daily for 10 days or until discharge, if sooner. In practice, the median duration of azithromycin was 6 days (IQR: 3 to 9 days). This was open-label for participants and local health workers, but study investigators were blinded to the outcomes.

Baseline characteristics included mean age 65 years (SD +/–15) with 58% being <70, 23% 70 to 80 and 19% >80 years. Ethnicity included 73% white, 14% BAME and 14% unknown; and 62% were men. Overall, 58% had a history of previous complications including diabetes (27%), heart disease (26%) or chronic lung disease (25%).

The median time since symptom onset was 8 days (IQR: 5 to 11 days) and time since admission to hospital was median 2 days (IQR: 1 to 4).

Other treatments of note included 46% using a corticosteroid, 20% using remdesivir and 17% using convalescent plasma. Supplemental oxygen was given to 76% and an additional 6% needed invasive mechanical ventilation.


The interim results were based on 73% of participants. Follow up will be complete by the end of December for the 27% who hadn’t reached this endpoint when this arm of the study was stopped.

Overall, 19% of participants died in each arm: 496 vs 997 in the azithromycin vs control arms respectively.

There was no difference in the primary endpoint of all-cause mortality at day 28 (RR 1.00; 95% CI 0.90 to 1.12), p=0.99). There were also no differences in use of mechanical ventilation, duration in hospital or in subgroup analyses (including age, sex, ethnicity, level of respiratory support, days since symptom onset, use of corticosteroids, and predicted 28-day mortality risk).

The pre-review paper summarising these results was also posted online, with much of the important detail included in supplementary material. [2] 

Statistical analysis

The study was based on a predetermined decision that a 20% reduction in 28-day mortality would be clinically significant. This was used to calculated numbers of participants needed to provide at least 90% power at two-sided p=0.01, adjusted by the background mortality reported in the study.

For example if mortality was 20% then the blinded Trail Steering Committee (TSC) calculated that the study would need 2000 participants in the experimental arm and 4000 in the control arm. As mortality in the study was lower, the TSC recommended these numbers increase to 2500 and 5000 respectively.

The trial protocol refers to the possibility of the trial stopping early due to benefit (needing “at least a 3 to 3·5 standard error reduction in mortality”.

However, the protocol does not appear to consider earlier options for lack of benefit.


Few people will be surprised at these results as three smaller RCTs have already published similar outcomes. [3, 4, 5]

The time taken to find this lack of benefit though is disturbing and upsetting, together with the overall (accumulating) mortality.

More than 20,000 participants have now been enrolled in RECOVERY. Mortality rates suggest that more than 4,000 people will have died, and that perhaps fewer than 100 participants had their lives saved in the study (if enrolled to the early dexamethasone arm).

Although the study design was approved as acceptable eight months ago, the independent Data Monitoring Committee should arguably have a more active role in the event of no signal of benefit.

Allowing experimental arms to continue for so long while waiting for the primary endpoint of significant benefit is no longer acceptable. Including an early threshold for activity should be considered for the remaining study arms: tocilizumab, convalescent plasma, REGEN-COV2 (two monoclonal antibodies), aspirin, and colchicine.

This high over mortality in the ineffective arms – and the control arm – urgently raises questions about the stop/go thresholds for deciding whether or not there is sufficient signal of benefit for the ineffective treatments to continue.


  1. RECOVERY press statement. Statement from the Chief Investigators of the randomised evaluation of COVID-19 therapy (RECOVERY) trial on azithromycin. (14 December 2020). (PDF)
  2. Horby PW et al. Azithromycin in hospitalised patients with COVID-19 (RECOVERY): a randomised, controlled, open-label, platform trial. (14 December 2020). (main study) (supplementary material)
  3. Cavalcanti  AB et al. Hydroxychloroquine with or without azithromycin in mild-to-moderate Covid-19. N Engl J Med 2020; 383:2041-2052. DOI: 10.1056/NEJMoa20190142020. (19 November 2020).
  4. Furtado  RHM et al. Azithromycin in addition to standard of care versus standard of care alone in the treatment of patients admitted to the hospital with severe COVID-19 in Brazil (COALITION II): a randomised clinical trial. Lancet 2020; 396(10256): 959–67. (3 October 2020).
  5. Sekhavati  E  et al. Safety and effectiveness of azithromycin in patients with COVID-19: An open-label randomised trial. Int J Antimicrob Agents 2020; 56(4): 106143. (October 2020).

This report was first posted on 7 January 2021.

Links to other websites are current at date of posting but not maintained.