What supply signal analysis actually looks at

Drug supply signal analysis watches a local drug market, not people. It tracks which substances appear to be circulating in a county, what they are being called and what toxicology is showing, using only openly published material. It never looks at program participants, never identifies buyers or sellers and never goes to enforcement; its output is a warning for public health, harm reduction and clinicians.

Of everything in this program, this is the component most likely to be misread, and the misreading is reasonable. “Social media analysis” in a program involving patients sounds like monitoring patients. It is not what this is, and the distinction is structural rather than a policy setting.

What does drug supply signal analysis look at?

The unit of analysis is a substance in a place. Not a person, not a caseload, not a set of accounts. The output is of the form: “this appears to be circulating in this county, it is being called these names, here is what toxicology is showing.” Nobody is identified because nobody is being looked for.

Why monitor the local drug supply?

Because the composition of a local supply changes faster than any reporting cycle. Carfentanil deaths recorded by CDC surveillance rose roughly sevenfold in a year, from 29 in the first half of 2023 to 238 in the first half of 2024. An adulterant can enter a supply, kill people for months, and appear in a published dataset a year later.

And the vocabulary changes deliberately. What a substance is called locally shifts constantly, often specifically so it is not recognized. A harm-reduction worker, an emergency physician and an epidemiologist in the same county can be describing the same thing in three different words. Tracking the vocabulary is what lets those three be connected.

What does supply signal analysis never do?

• It does not analyze the accounts of people in the program. Participants are not searched for, not matched to accounts, not looked at.
• It does not identify sellers or buyers. The output is vocabulary and geography in aggregate; there is no attribution step, and building one is outside scope.
• It does not access private communications and does not create accounts to reach closed spaces. Only openly published material.
• It does not go to enforcement. The product is a warning routed to public health communication, harm-reduction services and clinicians.

How is a drug supply warning confirmed?

An open-source signal is a hypothesis, not a finding. Before anything is issued, it is checked against toxicology and overdose data — see emerging threat analysis. A vocabulary shift with no corroborating toxicology is noise, and issuing warnings from noise destroys the credibility that makes warnings work.

If this still makes you uneasy

That is a reasonable response to the category, and the right answer is not for us to reassure you. Put it to your counsel and to whatever community oversight exists locally, and narrow the deployment until you are comfortable. We would rather a component were scoped down than have it quietly become something this page says it is not.

More: the component page.

Frequently asked questions

What is the process of drug supply monitoring here?

It starts with openly published material that suggests a substance is circulating in a county and names what people are calling it. That signal is treated as a hypothesis. It is checked against toxicology and overdose data, and only a corroborated signal becomes a warning, which goes to public health communication, harm-reduction services and clinicians.

Does drug supply monitoring share information with the police?

No. The component does not go to enforcement. It does not identify sellers or buyers, has no attribution step, does not read private messages and does not create accounts to get into closed groups. The only product is a warning routed to public health, harm reduction and clinicians.

Why track the street names people use for drugs?

Because the names change on purpose, often so a substance is not recognized. A harm-reduction worker, an emergency physician and an epidemiologist in the same county can describe the same drug in three different words. Tracking the local vocabulary is what lets those three reports be connected to one threat before it shows up in a published dataset a year later.