Ahhhhhhomg okay.
So prevalence CAN BE DONE RIGHT in a disease like ME!
I found the CDC's four-year prevalence study for autism, started in 2014 and finished up just last year (a four-year study). Here's how they did it:
First, they connected with state and local governments in order to get permission to acquire all kinds of data. They did so in 11 different states in order to do their best to get data from different places in the country. Even so, they acknowledge that this can't represent the whole country for real, and so prevalence will be their best estimate.
Then, they pulled "is someone potentially autistic?" from numerous, numerous sources. Not just ICD code -- it included school reports, doctor's reports, hospital admissions -- and lookalike diagnoses -- along with actual diagnoses of autism.
Finally, they compiled vast swaths of data on each of these individuals and handed them off to trained, expert clinicians and said, "based on your best judgment, do this person's results mean they are autistic?"
This is probably the only way to do it short of actually interviewing millions of people in person.
I chose autism because there are a lot of similarities in terms of epidemiological issues:
- There is no biomarker for autism
- Symptoms present very differently, person to person
- They have numerous diagnostic criteria and there is argument over which are valid.
Good case study for us.