A few people have sent me a study from the Netherlands (" Post-COVID-19 condition in individuals infected with SARS-CoV-2 in autumn 2023 in the Netherlands: a prospective cohort study with pre- and post-infection data") that claims to show that almost no one developed Long COVID during the fall wave of 2023.
Obviously, this doesn't align with anecdotes. At this point, I've dropped out of the Stanford Long COVID clinic, but when I went for my last appointment earlier this month, they told me that they're still getting a steady stream of new patients.
It's no surprise, then, that this study suffers from numerous methodological flaws. A few I found:
1. Their method of classifying someone as "uninfected" is deeply flawed. They used an antibody test that is known to lose its effectiveness quickly, or even give false negatives.
2. But their method of classifying someone as "infected" was even worse. They used self-reported at home tests. People who had asymptomatic cases (which can still give you Long COVID symptoms) or who chose not to test at all (something dangerous lunatic and FDA bureaucrat Vinay Prasad has advocated) wouldn't show up in their infected group.
These two things taken together mean that it's highly probable many people in their control group actually had COVID, and therefore, it's no surprise that if you compare the two groups, you don't see much difference.
3. You can actually see strong evidence for this directly in their data: Their "uninfected" participants had higher rates of headaches, sore throat, fever, runny nose, and abdominal pain compared to the COVID "infected" group. This flat out doesn't make any sense.
4. They more or less arbitrarily made up their own system for defining if someone had Long COVID. They used a questionnaire that had not been validated, and then created point scale cut off points more or less on a whim.
If you want to see how bad this methodology is, you can notice that some of the people in their "uninfected" control actually reported having Long COVID. Again, this doesn't make any sense.
5. The cohort they used is not a representative sample. They targeted older individuals (60-85). It's challenging to distinguish Long COVID in this group given their age.
6. To make their data work, they threw out participants who couldn't be matched. Of course, these participants tended to be younger, adding to the selection issues.
7. Their methodology required the patients regularly follow up. The study authors admit the people sickest with Long COVID would, naturally, be less likely to follow up (they could be bed-bound, housebound, living in a dark room, etc. not filling out surveys).
8. There were a bunch of weird little mathematical quirks they called out - lacking computation power to correct for repeat measurements. Inability to match groups by number of positive tests. Issues with the number and types of vaccines people took.
When anecdotes conflict with data, the anecdotes are usually right.
Obviously, this doesn't align with anecdotes. At this point, I've dropped out of the Stanford Long COVID clinic, but when I went for my last appointment earlier this month, they told me that they're still getting a steady stream of new patients.
It's no surprise, then, that this study suffers from numerous methodological flaws. A few I found:
1. Their method of classifying someone as "uninfected" is deeply flawed. They used an antibody test that is known to lose its effectiveness quickly, or even give false negatives.
2. But their method of classifying someone as "infected" was even worse. They used self-reported at home tests. People who had asymptomatic cases (which can still give you Long COVID symptoms) or who chose not to test at all (something dangerous lunatic and FDA bureaucrat Vinay Prasad has advocated) wouldn't show up in their infected group.
These two things taken together mean that it's highly probable many people in their control group actually had COVID, and therefore, it's no surprise that if you compare the two groups, you don't see much difference.
3. You can actually see strong evidence for this directly in their data: Their "uninfected" participants had higher rates of headaches, sore throat, fever, runny nose, and abdominal pain compared to the COVID "infected" group. This flat out doesn't make any sense.
4. They more or less arbitrarily made up their own system for defining if someone had Long COVID. They used a questionnaire that had not been validated, and then created point scale cut off points more or less on a whim.
If you want to see how bad this methodology is, you can notice that some of the people in their "uninfected" control actually reported having Long COVID. Again, this doesn't make any sense.
5. The cohort they used is not a representative sample. They targeted older individuals (60-85). It's challenging to distinguish Long COVID in this group given their age.
6. To make their data work, they threw out participants who couldn't be matched. Of course, these participants tended to be younger, adding to the selection issues.
7. Their methodology required the patients regularly follow up. The study authors admit the people sickest with Long COVID would, naturally, be less likely to follow up (they could be bed-bound, housebound, living in a dark room, etc. not filling out surveys).
8. There were a bunch of weird little mathematical quirks they called out - lacking computation power to correct for repeat measurements. Inability to match groups by number of positive tests. Issues with the number and types of vaccines people took.
When anecdotes conflict with data, the anecdotes are usually right.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
