What does "efficacy" mean for a COVID-19 vaccine?

This will be an unusual post, I apologize already by my uncertainty. I do want people to let me know where I may be missing something. It relates to the "efficacy" expected of the COVID-19 vaccines. I tried to search the EMA and FDA guidelines for approvals of vaccines. I also tried to find information on ClinicalTrials.gov but the answer appears elusive. The FDA states that they will approve a vaccine with 50% efficacy.

• To ensure that a widely deployed COVID-19 vaccine is effective, the primary efficacy endpoint point estimate for a placebo-controlled efficacy trial should be at least 50%, and the statistical success criterion should be that the lower bound of the appropriately alpha-adjusted confidence interval around the primary efficacy endpoint point estimate is >30%. (From FDA guidance to the Industry)

The media and various pundits are speculating when the first vaccine will get approval and estimations are for late this year to sometime next year. But here is the question. What does 50% efficacy mean for a vaccine? Again, from the FDA document:

• Either laboratory-confirmed COVID-19 or laboratory-confirmed SARS-CoV-2 infection is an acceptable primary endpoint for a COVID-19 vaccine efficacy trial. o Acute cases of COVID-19 should be virologically confirmed (e.g., by RT-PCR). 
• SARS-CoV-2 infection, including asymptomatic infection, can be monitored for and confirmed either by virologic methods or by serologic methods evaluating antibodies to SARS-CoV-2 antigens not included in the vaccine. 

It seems that the protocols indicate that the number of infections will be compared across groups. The timepoint valid for the study starts around 2 weeks after the last (or second) injection. So, if anyone gets COVID prior to the six weeks from the start of the trial (dosis are given 4 weeks apart), that infection will not count for the vaccine or placebo (a modified intent to treat regimen). If "efficacy" is a 50% reduction, you could get in a short period of time lots of infections in placebo and reach the statistical endpoint. But it seems that many expect that to happen very quickly! What do you think if the endpoint is reached at 3 months? Do you believe that the vaccine is "efficacious" if the period of protection shown for registration is only 3 months? The trials say that efficacy is up to one year. But, don't you want to show if the reduction in infections lasts one year PRIOR to the vaccine getting approval? Do you think that the public will understand that the trials only show 3 months of data, for example? The opposite could happen too if the trial sites are in areas with very low infection rates and it takes a very long time to reach the 50% reduction. The incidence rate of infections will determine the "duration" of the study, right?

So, let's assume that a vaccine trial reaches the statistical significance efficacy of 50% reduction in the number of infections by 3 months after the second injection. Let's assume that safety is not a problem. What happens if by month 4 or 5 there is significant loss of protection and vaccinated patients start to get infected? Do you think it is acceptable to have a vaccine and not know the duration of protection? Are current circumstances so exceptional that we are willing to take a risk? And this is if we observe only a 50% reduction. Is the enough?

As you can see, the elusiveness of this guidance troubles me. The media and the experts have not been clear in their communications of what the vaccine can or should achieve. One of the things that I learnt is you cannot buy time. You can enroll more patients, you can enroll fast, you can spend a lot of time and effort to get fast and quality enrollment. But you cannot buy the duration of the trial. If the protocols in ClinicalTrials.gov indicate up to one year, it can be considered as possibly misleading if the achievement of statistical significant 50% reduction is not time dependent.

I am not trying to be difficult. I am just trying to understand what is being done so I can better evaluate the vaccines once approved.

If anyone knows more how this will work, please, let me know.

Since I wrote this yesterday, Moderna has disclosed the Clinical Study protocol for their vaccine. Some of the questions that I have are covered in the protocol. The situation as described in the protocol actually reinforces my concerns.

The study says that the primary outcome, or vaccine efficacy (VE) is based on achieving BOTH outcomes, virological positivity AND symptoms. Virological positivity ALONE is only a secondary outcome measure. So you could have lots of positive cases with the vaccine without symptoms and only the symptoms result in achieving the primary outcome.

Moderna also assumes that by reaching 151 "positive" cases you could show a 60% "VE". That is, roughly 105 patients on placebo with both virological confirmation and symptoms versus roughly 46 patients on the vaccine. Don't hold me to the exact numbers, this is a back of the envelope quick calculation.

They will test if they reach significance at earlier time points. I guess it will depend on the infection rate. They assume a very low rate overall of .75%.

The 151 number is assumed to be reached by 10 months. So we will not know if there is a time dependence on the efficacy of this vaccine. We will only know that it eventually reached a 60% VE by the time of registration. Which could be 3 months on vaccine, 5 months or 10 months. We will not know if we have to retake it or when to retake it. We won't know how long the protection lasts either.

I don't know if this is common for vaccines, but it is certainly provoking some discomfort in me.