Atlas
In the Atlas, pick a condition and see where and for whom it goes uncoded: a state map, coding by age, sex, and race and ethnicity, what else is in the record, and for Long COVID, what finding patients earlier could cost.
With the default assumptions, a two-step screen costs $116 in clinician review time for each later-coded patient it finds, and chart review of every flag costs $233, each finding patients about 55 days before their Long COVID code.
Setting: 10,000 patients who had COVID-19 are scored by the model at day 90, and the top 5% (500 patients) are flagged. A later-coded patient is one who receives the Long COVID code later on.
$116 per later-coded patient found
Flagged patients answer a symptom questionnaire first, and a clinician reviews the 200 who pass. That finds 17.2 later-coded patients for $2,000 of clinician time, about 60 days before their code, if 80% of later-coded patients pass the questionnaire.
$233 per later-coded patient found
A clinician reviews all 500 flagged patients. That finds 21.5 later-coded patients for $5,000 of clinician time, about 60 days before their code.
Before their first code, later-coded patients have about $900 more in visits and tests than similar patients without the code. If earlier recognition avoids part of that, chart review of every flag avoids an estimated $3,193 against $5,000 of review cost, and the two-step screen avoids an estimated $2,554 against $2,000, per 10,000 patients screened.