repurposed from my comment on a FB post on an article criticizing all antidepressants as basically placebos
epistemic status: kind of dreading comments on this because it’s not well phrased, but honing it is too low a priority. Every time you criticize insufficient caveating an angel loses its wings.
medical studies are ~only concerned with the median person. Any unusual success or failure is written off as noise, instead of replicable variability. As conditions get defined they narrow that to “median person with condition X” rather than “median person”, but this only helps if you are carving reality at its joints.
E.g. most studies that find no effect in vitamin D don’t look at starting value of D, or if it went up with treatment. They’re examining the act of a random person taking Vitamin D, rather than trying to inform a model about what constitutes scarcity.
I’m pretty convinced that depression is a set of symptoms with multiple causes, like fever. If we had antibiotics and antivirals to treat fever but not the concept of bacteria or viruses, we might see results pretty similar to what we see in anti-depressants. Sometimes they work, sometimes they don’t, also lots of people get better without either, it seems like kind of a crapshoot.
But barring allergies (and more recently, evolved immunity), people respond pretty similarly to broad-spectrum antibiotics (but even then, some infections require narrower prescriptions). The brain is much more complicated. If you have a condition that’s inherently noisy (due to regression to the mean), and a given treatment is very good but only for 5% of people, that can easily get lost in the noise. Obviously you’d rather have a treatment that works for most people, but if it doesn’t exist and the condition is bad enough, lots of problems are worth trying 20 meds until one works.
This is especially nasty when you look at FDA standards, which are ~that a new drug must have greater (median) efficacy or fewer (median) side effects than existing drugs. This completely ignores the fact both of those can vary a lot per person, such that a given drug is worse for the mythical median person but much better for some percent of the population, who may not even be able to take the existing drugs.
With depression you additionally have issues of severity. Seems entirely plausible that the drugs work great for severe depression but are lost in the noise for mild depression (my personal experience is wellbutrin is great even for mild depression, but it’s also treating my nerve damage, and nothing helps depression like reducing chronic pain).
Additionally, I haven’t checked these studies in particular, but many tests for depression aren’t very good or sensitive.
repurposed from my comment on a FB post on an article criticizing all antidepressants as basically placebos
epistemic status: kind of dreading comments on this because it’s not well phrased, but honing it is too low a priority. Every time you criticize insufficient caveating an angel loses its wings.
medical studies are ~only concerned with the median person. Any unusual success or failure is written off as noise, instead of replicable variability. As conditions get defined they narrow that to “median person with condition X” rather than “median person”, but this only helps if you are carving reality at its joints.
E.g. most studies that find no effect in vitamin D don’t look at starting value of D, or if it went up with treatment. They’re examining the act of a random person taking Vitamin D, rather than trying to inform a model about what constitutes scarcity.
I’m pretty convinced that depression is a set of symptoms with multiple causes, like fever. If we had antibiotics and antivirals to treat fever but not the concept of bacteria or viruses, we might see results pretty similar to what we see in anti-depressants. Sometimes they work, sometimes they don’t, also lots of people get better without either, it seems like kind of a crapshoot.
But barring allergies (and more recently, evolved immunity), people respond pretty similarly to broad-spectrum antibiotics (but even then, some infections require narrower prescriptions). The brain is much more complicated. If you have a condition that’s inherently noisy (due to regression to the mean), and a given treatment is very good but only for 5% of people, that can easily get lost in the noise. Obviously you’d rather have a treatment that works for most people, but if it doesn’t exist and the condition is bad enough, lots of problems are worth trying 20 meds until one works.
This is especially nasty when you look at FDA standards, which are ~that a new drug must have greater (median) efficacy or fewer (median) side effects than existing drugs. This completely ignores the fact both of those can vary a lot per person, such that a given drug is worse for the mythical median person but much better for some percent of the population, who may not even be able to take the existing drugs.
With depression you additionally have issues of severity. Seems entirely plausible that the drugs work great for severe depression but are lost in the noise for mild depression (my personal experience is wellbutrin is great even for mild depression, but it’s also treating my nerve damage, and nothing helps depression like reducing chronic pain).