Union health minister Harsh Vardhan holds up a vial of Covaxin, AIIMS Delhi, January 16, 2021. Photo: Reuters/Adnan Abidi/File Photo
New Delhi: Even as its Brazil deal has gone belly-up, with the company staring at two criminal investigations into its attempt to import Covaxin into the South American country, Bharat Biotech has released the preprint paper describing the phase 3 trials of the beleaguered vaccine.
The company said in a statement that Covaxin demonstrated an efficacy of 77.8% in the trials.
“I am delighted to note that Covaxin developed by ICMR and BBIL, under an effective public-private partnership, has demonstrated an overall efficacy of 77.8% in India’s largest COVID-19 phase 3 clinical trial thus far,” Indian Council of Medical Research (ICMR) director-general Dr Balram Bhargava told PTI. ICMR developed and tested the vaccine along with Bharat Biotech.
Bharat Biotech has been receiving flak at home for not sharing the full data from its phase 3 trials as and when they become available, even though Covaxin has been part of India’s COVID-19 vaccination drive for more than five months. The new preprint paper, uploaded to the medRxiv repository, corrects this failure to some extent.
Overall findings
According to the paper, Covaxin is:
- 77.8% efficacious against symptomatic COVID-19 two weeks after the second dose
- 93.4% efficacious against severe COVID-19
- 65.2% efficacious against symptomatic COVID-19 caused by the delta variant (B.1.617.2)
The phase 3 trials were a multi-centre affair with 24,419 participants aged 18 to 98 years. Bharat Biotech and the Indian Council of Medical Research recruited them between November 16, 2020, and January 7, 2021. (The Drug Controller General granted ’emergency use’ approval for Covaxin on January 3, 2021.)
According to one of the trial’s registrations, its administrators could perform an interim analysis of the data at two intervals: once there were 43 and then 86 cases of COVID-19 among the trial participants. The efficacy analysis – reported in the new preprint paper – is based on the final threshold of 130 cases.
Originally, the trial enrolled 25,798 participants. They were divided roughly in a 1:1 ratio to the treatment and placebo groups, using the double-blind method (i.e. neither the participants nor the trial administrators knew which participant was in which group). Participants in the treatment arm received two doses of Covaxin 28 days apart, as intramuscular doses.
The primary outcome was the number of symptomatic COVID-19 cases two weeks after the second dose. As Dr Jammi Nagaraj Rao explained in an older article:
The virus’s ‘attack rate’ in the placebo group is effectively the ‘background rate’, or the infection rate we infer from real-world data. The vaccine’s efficacy is the percentage by which the ‘attack rate’ in the vaccine group is lower, weighted by the extent to which this effect is the result of the vaccine.
Based on this, the researchers estimated the efficacy to be 77.8%. The confidence interval (CI) denotes the range across which a measured value can vary, and the 95% CI – the ‘95%’ being a quality cut-off – for Covaxin’s efficacy was 65.2% to 86.4%.
Against asymptomatic COVID-19, the efficacy was 63.6%, with a 95% CI of 29% to 82.4%. And against symptomatic infections of the delta variant, Covaxin reportedly has an efficacy of 65.2% with a 95% CI of 33.1% to 83%.
The overall efficacy decreased to 67.8% for those recipients aged 60 years or older; the 95% CI was 8% to 90%. The efficacy increased to 79.4% for those aged 59 years or younger; the 95% CI was 66% to 88.2%.
And Covaxin’s 93.4% efficacy against severe COVID-19 had a 95% CI of 57.1% to 99.8%.
A problem in the CIs
Note that all these CIs are considered to be wide. The ones for Covaxin’s efficacy against asymptomatic COVID-19, against the symptomatic infections of the delta variant and in 60+ year-olds are extremely wide. It is the same as Bharat Biotech claiming Covaxin’s efficacy in 60+ people, against symptomatic COVID-19, is 8-99% with 95% confidence.
This is an indication that the underlying data was not ‘good enough’ to be able to tighten the measurements.
In early March 2021, Bharat Biotech had released some preliminary results from the phase 3 trials, based on 43 ‘events’ – i.e. 43 cases of COVID-19 among its 24,419 participants. It claimed that Covaxin was 80.6% efficacious. Dr Rao had cautioned at the time that this number may not be reliable because it was based on such a small number of ‘events’, yielding a 95% CI of 56.4% to 91.3%.
The new range is somewhat narrower, but still wide; it should ideally span about 10% points.
Safety profile analysis
The preprint paper also reports that 12% of Covaxin recipients reported moderate side-effects, 0.5% reported serious side-effects, and that there were no vaccine-related deaths.
Specifically, 99 participants reported severe adverse events (AEs) – 39 in the treatment arm and 60 in the placebo arm. The researchers say in their paper that they will monitor the participants’ health for up to a year after the date of the second dose.
- Local injection pain after first dose: 3.04% of treatment arm, 2.78% of placebo arm
- Local injection pain after second dose: 1.81% of treatment arm, 1.62% of placebo arm
- Most common solicited adverse events (i.e. ones that the clinicians asked about): headache, pyrexia, fatigue and myalgia; less than 1% in both arms
- Mild AEs: 11.2% in treatment, 10.8% in placebo
- Moderate AEs: 0.8% in treatment, 1.1% in placebo
- Severe AEs: 0.3% in treatment, 0.4% in placebo
Some missing info
Note that the preprint paper states that the safety profile analysis is based on all the 25,798 participants who originally enrolled for the trial, and that the final analysis reports a cohort size of 24,419 participants. This is considered the ethically right thing to do.
However, this also means 1,379 participants were excluded from the efficacy analysis – likely because the dropped out, died or withdrew their consent to participate during the trial itself. The paper does not elaborate on this count, although it presents a flow diagram (see below).
