5 ms·

> Could it have something to do with being locked up for the last 6 months or worry about ones future career maybe? So you're saying the study is so shoddy tha

> Could it have something to do with being locked up for the last 6 months or worry about ones future career maybe? So you're saying the study is so shoddy that they didn't have a control group of people who went through the same lockdowns but didn't get COVID? Otherwise, your comment doesn't make sense.

rriepe 6y ago
The proper control group would be people with COVID who didn't go through the lockdowns.
- happytoexplain 6y ago
  They both seem like useful control groups - can you explain the difference to an amateur on statistics?
  - rriepe 6y ago
    Covid and the lockdowns both might be causing the effect. So you have one group that has just Covid and then one group that has both. Then you compare the outcomes and see if the just-Covid group has the effect too. But my comment was kinda tongue-in-cheek: How do you even find people not affected by this doom-n-gloom global news story?
- tablespoon 6y ago
  > The proper control group would be people with COVID who didn't go through the lockdowns. No, they were studying the effects of COVID, not the effects of lockdowns. Your suggestion is controlling for the wrong variable.
- kkoncevicius 6y ago
  Best control group would be people who went through the lockdowns, had flu but suspected themselves to have COVID.
  - rriepe 6y ago
    I like this one too.
taldo 6y ago
Their control group was people that had influenza and other respiratory conditions in the same timeframe. The hazard ratios for COVID survivors (over influenza or others) were in the 1-2% range. Meh.
- kazinator 6y ago
  That's not a credible control, because influenza and respiratory conditions are not a plausible placebo for Covid-19. The subjects knew that they are not having Covid-19, which makes them different from subjects which got Covid-19 and knew it.
  - Barrin92 6y ago
    I'm not really sure what you're trying to say, that if you control for covid-19, the effect of covid-19 goes away? Duh What you want to control for is the lockdown effects or general disease stress, not covid-19 itself. That's what you're trying to measure.
    - necovek 6y ago
      I think the GP has a point: ideally, you'd have unaware Covid19 patients as a control group, to see if this is psychological or physiological. Those aware they had Covid19 will fear all the known-unknowns around Covid19. Those who don't think they had it wouldn't have stressed about it at all.
  - c6401 6y ago
    Wow people do incredible mental hymnastics to prove themselves they are safe and all these studies are flawed. No matter how much effort have been put into research it will never be good enough.
kazinator 6y ago
You cannot control for that. You can't give a control group a "placebo illness" such that they don't know whether it's Covid-19 or not (and make it double-blind so that the people administering this illness also don't know). The people who overcame Covid-19 had an experience that members of a control group who did not overcome Covid-19 do not have. That's an obvious source of bias, which is inherent to the study. What would be credible would be data about people who had Covid-19 but did not know it (due to mild symptoms and no hospitalization), showing that those people have mental problems compared to a control group of those who didn't have Covid-19. However, if there is a real physiological effect there, it could be tied to the severity of the illness: i.e. how much damage it caused to the body. In that case, those secondary effects could be expected to be next to nonexistent in the mild group.
- loopz 6y ago
  You can never really rely on just one study anyways. You then need metastudies and comparative studies where such flaws get weeded out or corrected for. So there will be significant lag in certainty, thus not suitable for much of decisionmaking now.
- necovek 6y ago
  The effect might not even be physiological yet still be this prominent: having COVID-19 makes you doubly scared for your loved-ones, worry about your perception in the society (was I not careful enough?), worry about lingering side-effects, worry about missing extended time from work, etc. While this depends on the individual, you can notice effects of increased stress in large numbers. Your suggestion of another control group with Covid-19-but-no-symptoms would help answer that question too, though ideally we'd want those who had Covid19 but didn't know it at all. But anyway, this paper is only establishing correlation: getting to causation in medicine is the harder task and takes years or decades and even then it's usually inconclusive!

rriepe 6y ago

The proper control group would be people with COVID who didn't go through the lockdowns.

happytoexplain 6y ago
They both seem like useful control groups - can you explain the difference to an amateur on statistics?
- rriepe 6y ago
  Covid and the lockdowns both might be causing the effect. So you have one group that has just Covid and then one group that has both. Then you compare the outcomes and see if the just-Covid group has the effect too. But my comment was kinda tongue-in-cheek: How do you even find people not affected by this doom-n-gloom global news story?
tablespoon 6y ago
> The proper control group would be people with COVID who didn't go through the lockdowns. No, they were studying the effects of COVID, not the effects of lockdowns. Your suggestion is controlling for the wrong variable.
kkoncevicius 6y ago
Best control group would be people who went through the lockdowns, had flu but suspected themselves to have COVID.
- rriepe 6y ago
  I like this one too.

taldo 6y ago

Their control group was people that had influenza and other respiratory conditions in the same timeframe. The hazard ratios for COVID survivors (over influenza or others) were in the 1-2% range. Meh.

kazinator 6y ago
That's not a credible control, because influenza and respiratory conditions are not a plausible placebo for Covid-19. The subjects knew that they are not having Covid-19, which makes them different from subjects which got Covid-19 and knew it.
- Barrin92 6y ago
  I'm not really sure what you're trying to say, that if you control for covid-19, the effect of covid-19 goes away? Duh What you want to control for is the lockdown effects or general disease stress, not covid-19 itself. That's what you're trying to measure.
  - necovek 6y ago
    I think the GP has a point: ideally, you'd have unaware Covid19 patients as a control group, to see if this is psychological or physiological. Those aware they had Covid19 will fear all the known-unknowns around Covid19. Those who don't think they had it wouldn't have stressed about it at all.
- c6401 6y ago
  Wow people do incredible mental hymnastics to prove themselves they are safe and all these studies are flawed. No matter how much effort have been put into research it will never be good enough.

kazinator 6y ago

You cannot control for that. You can't give a control group a "placebo illness" such that they don't know whether it's Covid-19 or not (and make it double-blind so that the people administering this illness also don't know). The people who overcame Covid-19 had an experience that members of a control group who did not overcome Covid-19 do not have. That's an obvious source of bias, which is inherent to the study. What would be credible would be data about people who had Covid-19 but did not know it (due to mild symptoms and no hospitalization), showing that those people have mental problems compared to a control group of those who didn't have Covid-19. However, if there is a real physiological effect there, it could be tied to the severity of the illness: i.e. how much damage it caused to the body. In that case, those secondary effects could be expected to be next to nonexistent in the mild group.

loopz 6y ago
You can never really rely on just one study anyways. You then need metastudies and comparative studies where such flaws get weeded out or corrected for. So there will be significant lag in certainty, thus not suitable for much of decisionmaking now.
necovek 6y ago
The effect might not even be physiological yet still be this prominent: having COVID-19 makes you doubly scared for your loved-ones, worry about your perception in the society (was I not careful enough?), worry about lingering side-effects, worry about missing extended time from work, etc. While this depends on the individual, you can notice effects of increased stress in large numbers. Your suggestion of another control group with Covid-19-but-no-symptoms would help answer that question too, though ideally we'd want those who had Covid19 but didn't know it at all. But anyway, this paper is only establishing correlation: getting to causation in medicine is the harder task and takes years or decades and even then it's usually inconclusive!