Yesterday, Florida followed Norway by not recommending the Covid vaccine for children. The CDC recommends them.

What does the data say? As public health scientists, we must be honest both with what we know and what we do not know.

The emergency use authorization for the Pfizer-BioNTech mRNA vaccine for children was based on two randomized clinical trials for ages 5-11 and 12-15, respectively, with a total of 4,528 subjects. In both trials, there was a reduction in mild Covid infections during the two months following the second dose, with the vaccine efficacy in the 68% and 98% range for the younger children and somewhere between 75% and 100% for the 12-15-year-olds (95% confidence intervals).

What these numbers mean is that if the true value is, e.g., 90%, and if 100 children would have been infected without vaccination, then 90 of them will avoid the infection if vaccinated, while 10 children will still get infected despite being vaccinated.

A vaccine that only prevents mild disease is of little use, so what about serious disease, hospitalizations and deaths? There were zero such events among those who received the vaccine. There were also zero such events among those who received a placebo.

Hence, from the randomized trials we do not know if the Covid vaccines prevent hospitalizations and death among children. Neither do they tell us whether the protection against mild infection lasts longer than two months, or whether the vaccine reduces transmission.

With limited information from the randomized trials, we must turn to observational studies and we now have one. In New York State, 23% of children ages 5-11 and 62% of children ages 12-17 had been fully vaccinated by the end of January 2022. These 1.2 million vaccinated children were studied from November 29 to January 30, comparing them to the unvaccinated children in the State. Here is what we learned from that study, with all risk estimates based on 95% confidence intervals.

1. The New York study confirms the results from the randomized trials. The vaccine reduces short-term infection risk. During the first two weeks after the second dose vaccine efficacy against infection is in the 62%-68% range for 5–11-year-olds and in the 71%-81% range for 12–17-year-old children.
2. The protection against infection wanes rapidly. In the fifth week after vaccination, the vaccine efficacy against infection is in the 8%-16% range for 5-11-year-old children and in the 48%-63% range for the older ones. In the seventh week after vaccination, vaccine efficacy dropped further, to the 18%-65% range for the 12-17-year-olds. This is consistent with the rapid waning in protection that we have seen among adults, although the decline seems more rapid for children.
   
3. For the 5-11-year-olds, vaccine efficacy is negative during the seventh week after the second dose, with the unvaccinated having a lower risk of infection in the 29% -56% range. How can this be? A likely explanation is that the unvaccinated children got infected earlier than the vaccinated ones, and once the protection has worn off, the vaccinated children are at higher risk than the unvaccinated ones who have now acquired natural immunity. That is, the vaccine simply postponed infections by a few weeks or months.
   
4. How about deaths from Covid? That is what really matters. Unfortunately, the New York study does not present mortality data. Why? Over the two years of the pandemic, the survival rate for New Yorkers ages 0-19 is 99.999%. Despite over 3 million children, there may not have been enough Covid deaths during the two-month study period to determine vaccine efficacy against mortality. It would still have been useful to tally the numbers, but the study authors did not do so.
   
5. For hospitalization, the study reports that vaccine efficacy is higher than for infection, and while that protection also wanes over time, the decline is slower than for infections. The numbers reported mean that by vaccinating 365,502 children ages 5-11, an estimated 90 hospitalizations were prevented. This would mean that in order to prevent one hospitalization, one must vaccinate 4,047 children. The corresponding number is 1,235 for children ages 12-17.
   
   These numbers are difficult to properly interpret for four reasons. (i) They are based on a two-month period, and the vaccines have additional benefits outside that time window. (ii) They compare vaccinated children with unvaccinated children with or without natural immunity from prior Covid infection. This will underestimate the vaccine benefits for children without a prior infection while overestimating the benefits for those with natural immunity. (iii) They include both hospitalizations that are due to Covid and hospitalizations for other causes with a concurrent unrelated mild Covid infection. Even if the vaccine had zero efficacy at preventing hospitalization due to Covid, the efficacy against mild Covid infection would ensure that the study reported good efficacy against hospitalization. That the reported vaccine efficacy is higher for hospitalization than for infections indicates that there is at least some efficacy for the former, but it is impossible to properly estimate the level of efficacy without data that distinguishes hospitalizations due to and with Covid. (iv) The study was conducted during a large wave of infections, which has since declined. The benefits are less during the lower transmission period that we have now entered.

When deciding whether to vaccinate a child, we must also consider known and potential adverse reactions. From the CDC’s Vaccine Safety Datalink we know that the Pfizer and Moderna vaccines can cause myocarditis among adolescents and young adults. Current risk estimates are in the range of one myocarditis for every 3,000 or 8,000 vaccinated adolescents and young men. Women have lower risk. There may also be additional still unknown adverse reactions.

The Covid vaccine has been widely used for children without solid information about its efficacy on hospitalizations and deaths, and without the ability to conduct a proper benefit-risk evaluation. The recent observational study from New York State adds a few important pieces to the puzzle, but we still do not know whether the benefits outweigh the risks.

For older people who have not yet had Covid it makes sense to get vaccinated. While there may be unknown low-risk adverse reactions, the large reduction in mortality risk far outweighs any such risks. For children, the mortality risk is very small and the known and any still unknown risks from adverse reactions may outweigh the benefits at reducing hospitalizations and death from Covid, which are unfortunately still unknown.