The term ”survivor bias” is usually used metaphorically, but in this case it applies literally.
Putting in some caveats about the risks associated with infection is not enough. This is a garbage analysis that should never have been accepted for publication.
I'm fully vaccinated as is my whole family but you're the one showing bias here. The data is the data. Obviously the data set is only recovered patients. That has no bearing on the analysis. It's not a recommendation to go get infected or anything ridiculous like that. It's not a recommendation of anything. It's just data.
That's the wrong analogy. Seat belts only work while you're wearing them. They don't prevent crashes. Imagine if the study found the opposite conclusion. That vaccines produced a stronger immune response than a live infection. We'd have a whole different take if that were the result so it's extremely valuable to have an answer.
You’re missing my point. We know nothing about the sample that was excluded (the people who died). Were the survivors already more prone to wearing seat belts, or did the accident change their behavior?
In the Covid case, is the sample more weighted towards people who are more naturally prone to have a robust immune response (in contrast to the missing sample, I.e. everyone who died), or is the disease alone responsible for the immune response?
It’s not an apples-to-apples comparison, so no valid conclusions can be drawn.
That's still irrelevant. They asked a specific question and got an answer. That answer adds to our knowledge. That's it. We already have data telling us how many people who contract covid will have life-threatening symptoms without treatment.
What does it add to our knowledge? That some sample of people have more antibodies than a non-equivalent sample? I could have told you that without a scientific study.
That same sample of people who had the illness may have had more, fewer or equal antibodies if they had not gotten sick and instead got vaccinated. This study does not tell us anything about which is the more likely possibility.
No you couldn't. You could have guessed but you wouldn't know. Now we have numbers. We know the level of immune response and the distribution. Like I said, imagine the opposite. If convalescence produced a weaker response than vaccines. That's equally possible and would have lead to some serious policy shift. Now we have a clear answer and can make informed decisions.
I couldn’t have guessed that we could find two samples of people such that antibodies in one group exceeded antibodies in the other? That is true for any non-uniform data set.
As for your other point, I also maintain that the opposite finding would have told us equally little. Perhaps they had lower antibodies because they had less virulent variants/cases on average, which is why none of them died - again, “survivor bias”. It would still tell us nothing about how their immune response would compare between the case where they got sick and the case where they got vaccinated.
Edit: here is a simple challenge that can resolve this discussion in your favor. If you can tell me definitively what the expected antibody level for the survivor dataset was prior to this study, I will concede that you are right and this is a meaningful analysis.
I'm confused. If I could tell you want the expected level was before the study, then you'd be right not me. My point is that we didn't know for certain until the study was done. We could reasonably predict it would be "some" but not that if it was more or less than a vaccine response.
Scientists can’t develop a null hypothesis about the impact of the most studied virus in the history of humanity? If you have no null hypothesis then you shouldn’t be drawing conclusions from data, period.
Anyone who reads this site regularly is probably aware of the problems with non-registered data analyses in academia (often referred to as the “replication crisis”).
We should be incredibly skeptical of any non-registered analysis, even when sample controls are used. When the study is unregistered AND the sample is uncontrolled the default posture should way beyond skepticism - it should be considered meaningless until reproduced by a better-structured analysis.
I don't see survivor bias here at all. This is simply a study that caters to a rather significant population who justifiably want to know what they stand to gain from this vaccine.
What this study has done is identified a sub population who stand to gain such minuscule benefit from vaccination that it would be unethical to include them in a study (most studies DO exclude previously infected individuals, huh wonder why?) Furthermore it is unscientific discrimination to penalize people when they could prove prior infection and/or immunity.
peterbonney's right for exactly the reasons that the public trying to reason its way through interpreting studies like this unaided is dangerous to public health.
This is an observational study. It's not possible for it to control for (among other things) the possibility that those who were infected and survived already possessed enhanced natural immunity due to some other cause. All we know is that people who get COVID-19 and don't die from it are less likely to become symptomatic with the delta variant than people who are inoculated, not whether or not they were always less likely to become symptomatic with the delta variant.
People who interpret this study to mean "I don't need the vaccine" are really rolling the dice on the possibility that there's an X-factor here to COVID-19 survival... And they don't have it.
I see they recovered. You're correct... One interpretation could be that a high percentage of X-factor was present in this study's recovered group. But since we don't know what the X-factor is, an unvaccinated person is rolling dice by assuming they have it.
Nothing about this study guarantees an unvaccinated individual survives a brush with COVID-19; it suggests they're likely to survive a second brush with it (and likelier still if they get vaccinated).
That’s not strictly true. There have been deaths, but this can be attributed to components in the vaccines that these unlucky people may have gotten at any time. There have been rare and unusual vaccine caused deaths.
In this study nobody died from vaccine. But the big thing thats not adjusted for is once you got Covid you will wear mask and won't go to restaurant, this is not adjusted for in this study.
Do you honestly think its wrong or uncommon to study a group of people who have experienced a disease? Perhaps in an effort to make them informed about their particular risks?
You haven't read the replied upthread where people use research like this to suggest "Corona parties" (like chickenpox parties) and how many people think it's safer to catch the virus than get the vaccine?
This article will absolutely not just be used by already-infected people trying to make a decision.
Its a problem for the main narrative but that is because the main narrative is brittle and stupid. I don't see why scientific news ought to be forced to cater to a particular political narrative. Let the main narrative fold this in along with whatever caveats they wish, leave the science publications out of it.
Putting in some caveats about the risks associated with infection is not enough. This is a garbage analysis that should never have been accepted for publication.