sentinel surveillance
a monitoring design that collects detailed data from a selected network of sites and projects national estimates from them — faster and richer than a census, and only as representative as the network.
when you read "an estimated 8.4 million drug-related ER visits", nobody counted 8.4 million visits. a few dozen hospitals were counted carefully, and the rest is statistical projection. that is not a flaw — it is a design, with known edges.
how it works
A sentinel system recruits sites — hospitals, clinics, labs — and gathers detailed data (often chart-level) that no national census could afford. Weights then project the sample to national totals. SAMHSA's DAWN system, the source of this site's ER-visit figures for psychiatric medications, ran on 60 contributing hospitals in 2024: part hand-picked high-priority sites, part sampled.
how to read its numbers
Two questions before quoting a sentinel estimate. First, what is outside the frame? For DAWN: VA, specialty, and psychiatric hospitals — a notable exclusion when the topic is psychiatric medication. Second, what does a case mean? DAWN counts a visit when chart reviewers judge a drug contributed — adverse reactions and withdrawal included, not just overdose. The estimate is real; its definition is doing quiet work.
somewhere to put it
free. anonymous. people who’ve been where you are 🤍
get Resolv Social — it’s freewant the deeper story? read prescribed to fail
questions
are sentinel estimates less trustworthy than surveys?
They answer different questions. A survey asks people; a sentinel network reads records at selected sites. Sentinel data is often more clinically detailed and less representative; the confidence intervals published with the estimates carry the sampling uncertainty, not the frame exclusions.
more from the glossary
resolv social is not a clinical product and does not diagnose, treat, or cure any condition. if you’re in crisis, call or text 988 (u.s.), 24/7, free.