Originally Posted by
Excargodog
No, severe cases are not a subset of mild cases, but that wasn’t the analysis. Severe cases are MOST DEFINITELY a subset of TOTAL cases and the null hypothesis - if you are claiming a decrease in severity of cases - involves the severity of the immunized versus the non-immunized CASES. Clearly, those who never got infected AT ALL were at no risk to have severe cases, be they immunized or unimmunized.
So among those AT ACTUAL RISK TO HAVE SEVERE CASES you got 30 out of 185 for the placebo group and 0 for 11 for the immunized group. Those are good results for the treatment group, don’t get me wrong, but the EXPECTATION based upon probability of severe case given infected is still only about 15%. The EXPECTATION for the eleven cases that got COVID notwithstanding the immunization was that they would have somewhat less than 2 cases. The data was that they had 0, and that’s good. But that’s still sparse data. The confidence interval for that data may have 100% as one of its bounds, but it’s WAY TOO early to say that it’s 100% effective in stopping severe cases.
Math in public but here we go...
No, there were another approximately 174 vaccine recipients who statistically did get EXPOSED but did NOT get severe covid (or any symptomatic covid, not sure if they tested for antibodies for asymptomatic covid).
Vaccine prevented about 30 cases of severe covid. Also prevented about 174 cases of mild covid, with 11 getting through.
We don't know or really care how many of the mild cases would have been one of the avoided severe cases.
Vaccine performance:
Out of 15,000 vaccine recipients; (this number doesn't matter, could have been 200 or could have been 200,000. 15K was selected to get enough exposure, quickly enough, given known covid prevelance in the trial population).
About 185 got sufficient exposure to get infected;
30 of those would have got severe covid (but didn't). Period. Stop. That's all we need to know for efficacy vs. severe covid, and it's good news. In reality the zero is likely not zero, but a very small fraction since there are people out there with compromised immunity, etc. But 30 prevented out of 185 exposed gets us where we need to be for practical purposes.
Additionally, 11 got mild covid.
Two cases prevented, as you say, would not be statistically significant. 30 cases vs. 28 cases would not be very significant. 30 cases vs. zero cases (or 0.001 if you like) is statistically significant.