AstraZeneca – Covishield delayed booster: Classic case of using bad science in the Lancet to make politically convenient decisions

“Modelling analyses showed an increase in vaccine efficacy after two standard doses from 55·1% (95% CI 33·0 to 69·9) with an interval of less than 6 weeks to 81·3% (60·3 to 91·2) with an interval of at least 12 weeks. A single standard dose had an efficacy against symptomatic COVID-19 in the first 90 days of 76·0% (59·3 to 85·9), yet provided no protection against asymptomatic infection (vaccine efficacy –17·2% [–248·6 to 60·6]).”

This one line in the Lancet paper that claimed delayed delivery of booster dose of the AstraZeneca vaccine increased its efficacy from 55% to 81% has been responsible for decisions by several countries to administer a booster dose 12 weeks after the first dose. But does this line stand up to reason or scrutiny? Is it backed by some hard data of value significant enough for countries to base their immunization schedule on? The unequivocal answer, as we shall see, is NO. Government after government has been utilizing this article to cover up for what the problem really is: Not enough vaccine for booster doses to be given in time. A marriage of bad science with political convenience for making bad decisions. The introduction to the article admits so much in its own words: “It offers much-needed evidence for the UK policy of extending the dosing interval to 12 weeks and for rapid mass-immunisation campaigns worldwide.” Clearly, the decision to offer delayed dosing of vaccine by the UK came first, and evidence followed subsequently.

The article starts with a disclaimer on its own limitations: they weren’t prospective trials by design and hence “could be biased”, they weren’t randomised, “only one of the four trials was double-blind, and the single-dose recipients were self-selected”, “baseline characteristics between the single-dose and two-dose cohorts were substantially different, with an older median age, higher proportion of men and non-white participants, and a smaller proportion of health or social care workers in the two-dose cohort than in the single-dose cohort”.

In short, the data was an incompatible hodge-podge of multiple trials of different natures clubbed together. While one UK trial excluded all people with co-morbidities, the second UK trial as well as the Brazil and South African trials included them. This defies accepted methodology of clubbing together data from trials that have been similarly conducted but avoiding it when methodology differs.

First, let us look at vaccine efficacy. The article claims that the two standard doses offer 63.1% efficacy as compared to low dose plus standard dose (80.7%). On every parameter, the first low and second standard dose regimen does better than two standard doses. But remarkably, no policy decision was taken by any country to offer low first dose plus second standard dose based on this data.

Now let us look more closely at the data that claims higher efficacy on delayed dosage: the absolute numbers in each group are not uniform. The groups are not standardized or randomized. The result is a wide discrepancy of 3890 volunteers in < 6 weeks group and 1293 in the > 12 weeks group. Absolute number of symptomatic positive patients is also low, with 35 in < 6 weeks group (0.9%) and 8 in >12 weeks group (0.6%) testing positive as compared to 2% and 3.3% in the control group respectively. Efficacy has been derived by comparing positives of 0.9% in test vs 2% in control (55.1%) and 0.6% in test vs 3.3% in control (81.3%). Radical decisions such as planning an entire country’s vaccination schedule are being made on test positives of single and double digit absolute numbers in the concerned groups.

The table above shows that efficacy of a booster dose was 55.1% if given at under six weeks and rose up to 80% if the booster was delayed to beyond 12 weeks. But the real kicker comes after this. The article then states, “A single standard dose of vaccine provided protection against primary symptomatic COVID-19 in the first 90 days with an efficacy of 76·0%, but was not efficacious against asymptomatic infection over the same time period. Efficacy of a single standard dose against any NAAT-positive infection was 63·9% from 22 days to 90 days, suggesting the potential for a substantial reduction in transmission, although these results are exploratory and require further investigation.

This is where the entire argument being set up in favour of delayed second dose falls apart completely. If the first dose has a standalone efficacy of 76% up to 90 days, it stands to reason that administration of second dose is only a booster that will raise efficacy above this 76%. In effect, what the Lancet paper is implying is that efficacy of a single dose is 76% but after administration of a booster at 4 to 6 weeks, it fell by over 20% to just 55%. Such an extraordinary claim unprecedented in medical history would have to stand on the back of extraordinary evidence. Which is just not there.

And have a look at the number of volunteers protected by a single dose at 61-90 days: the efficacy in preventing symptomatic COVID is just 31.6% but in preventing asymptomatic COVID is 67.6%. But it is still 79.4% for NAAT-positive (RTPCR-positive) COVID infections! In effect, the data in the above table implies that the vaccine performs poorly against preventing symptomatic COVID but offers good protection against asymptomatic COVID in the time window of 61-90 days. This is in stark contrast to the conclusions drawn from the article that the vaccine has low efficacy in protection against asymptomatic COVID but high efficacy against symptomatic COVID. Also, on clubbed up data, if the vaccine shows negative efficacy of -17.2% against asymptomatic COVID at 22 to 90 days, how does it suddenly shoot up to +67.6% in the time period of 91 to 120 days?

The severe discrepancies pointed out above show that either the data itself is seriously flawed and does not stand up to even a basic level of scrutiny, or that it is deliberately manipulated.

So why is the 55.1% efficacy number at 4-6 weeks in the picture at all? My cynical view is: to make the difference stand out. If the article were to state that a single dose offers 76% protection while a second dose administered at 12-20 weeks increases it to 81%, the benefit of a booster does not appear to be substantial (although a booster will undoubtedly be useful to prolong the duration and intensity of immunity). So why not throw in a straw man into the picture, and use it to show a huge benefit with delayed dose as compared to an earlier dose?

As the article stated upfront, the data used for it is questionable, not randomized, not standardized and potentially biased. It has some astonishing indirect implications, unprecedented perhaps in medical history, that two doses of a vaccine are worse than one, that administration of a booster at 4-6 weeks lowers efficacy of the first dose, that in some sub-groups the vaccine protects against asymptomatic infection but not against symptomatic infection. This is bad science, staring at us in the face. It was still published by The Lancet and taken as a gospel for making policy decisions by India and the UK. It is worrisome that politically convenient decisions on timing of administration of booster dose for COVID vaccination of more than one country are being based seemingly on the reputation of a journal instead of genuine science.

Update: it is proposed that the lowered second dose efficacy is due to use of the same chimp adenovirus in both doses. But this can at best explain why a second dose won’t work, it cannot explain why exposure to second antigenic stimulus 4-6 weeks after initial exposure brings down efficacy. And it again does not stand to reason that exposure to the same antigen three months later will increase efficacy.

Differing opinions are welcome as long as you keep it clean, so go for it :)