A Federal Register notice published August 20, 2026 proposes a significant change to the National Health Interview Survey. As part of a systemic redesign scheduled for 2028, the draft adult questionnaire includes a section titled “Alcohol, Nicotine, and Marijuana Use,” opening with:
“Do you currently use marijuana every day, most days, some days, or not at all?”
CBD-only products are excluded. Respondents who use cannabis would then be asked why — recreation, pain relief, nausea, anxiety or stress, sleep, or other medical purposes.
Written comments are due on or before October 19, 2026.
This site spends most of its time arguing that cannabis data collection is under-protected. So it is worth saying clearly: this proposal is, on the merits, how it should be done. The interesting part is why — and what that tells operators about everything else happening this month.
What NHIS Already Asks
This is not the survey’s first encounter with cannabis. The 2024 and current 2026 NHIS editions include a narrow question: “During the past 30 days, how often did you use marijuana or CBD products to help you fall asleep or stay asleep?”
That is sleep-specific and lumps CBD in with THC, which limits its analytical value considerably. The 2028 proposal separates the two, asks about general frequency, and adds the reason-for-use dimension.
The redesign notice also covers continuation of current questionnaire content and data collection operations through 2027, with the restructured instrument arriving in 2028. So this is a long runway, not an imminent change.
Why This Is the Good Version
NHIS is not a data broker product, an agency social listening contract, or a state registry. It is the principal source of health statistics on the U.S. civilian noninstitutionalized population, run by the National Center for Health Statistics and fielded by the Census Bureau through in-person household interviews.
Four properties make it materially different from most federal cannabis data collection:
Statutory confidentiality. NHIS responses are protected under Section 308(d) of the Public Health Service Act and the Confidential Information Protection and Statistical Efficiency Act. Information collected for statistical purposes may not be used for any other purpose, and identifiable data cannot be disclosed to enforcement or administrative agencies. This is a legal wall, not a policy preference, and violating it carries criminal penalties for the officials involved.
Statistical purpose limitation. The data exists to produce estimates. There is no operational use case — no watchlist, no enforcement referral, no adjudication that consults a respondent’s answers.
Voluntary participation with informed consent. Respondents are told what the survey is and can decline any question.
Public rulemaking. The reason you know this is being proposed is that the agency is required to publish it and take comment. You get sixty days to object. That is the entire point of the process, and it is the opposite of learning about a data collection after it exists.
Compare that to a state medical cannabis registry — which we have written about at length, including the structural absence of HIPAA protection and registries housed inside law enforcement agencies. Registry data identifies individuals, exists for operational purposes, and its access rules vary by state and shift with politics.
The CDC proposal collects less identifying information, under stronger legal protection, for a narrower purpose, with public notice, than the registry a patient already enrolled in.
The Contrast That Matters
Now put the month together.
August 20: CDC publishes a Federal Register notice proposing cannabis questions on a statistically protected, voluntary, consented household survey. Sixty days for public comment.
August 20: FDA posts a solicitation for a contractor to monitor cannabis conversations on TikTok, Reddit, and Instagram — filterable by inferred age group, sex, and persona, using AI modeling, with click-through to verbatim individual posts. Twenty-one days to respond, and no notice to any of the people whose posts are being indexed.
Same government, same fortnight, same subject. One collects less and protects it under criminal-penalty statutes. The other assembles an individually retrievable corpus in contractor custody with unknown retention.
The lesson is not “federal cannabis data collection is bad.” It is that collection method determines risk far more than collection subject does. A question asked openly, answered voluntarily, and firewalled by statute is a fundamentally different object than the same information scraped, inferred, and indexed.
That is the frame operators should apply to their own systems, and it is the argument to make when a regulator proposes a new reporting requirement: not whether data is collected, but under what purpose limitation and with what wall around it.
The Real Concern With the CDC Proposal
There is one, and it is not about the CDC’s handling.
Respondents do not know they are protected. A Census interviewer at the door asking a resident of a prohibition state whether they use marijuana, and why, will get inaccurate answers — not because the protection is weak, but because the respondent has no reason to believe it is strong. The consequence is undercount, concentrated precisely in the populations where accurate public health data matters most.
That is a survey design and communications problem worth raising in comments. The confidentiality assurance needs to be prominent, plain-language, and specific about the enforcement wall.
The medical-reason question is the sensitive one. Frequency of use is one thing. “I use cannabis for anxiety” is health information about a federally illegal substance, and it is the field most likely to be under-reported and most valuable if accurate. It deserves the strongest confidentiality framing in the instrument.
Neither concern argues against the questions. Both argue for saying so before October 19.
What Operators Should Actually Do
1. Comment by October 19. Trade associations should file. Useful asks: prominent confidentiality assurance language, plain-language explanation of the enforcement firewall, and care in how the medical-reason response options are worded so patients are not forced into a false binary between “medical” and “just to get high.”
2. Use the resulting data. Better national estimates of medical versus recreational use, by condition, is exactly the evidence base the industry has lacked in rescheduling and state legislative fights. Support the collection because you will want to cite it.
3. Adopt purpose limitation as an internal standard. The thing that makes NHIS defensible is that the data cannot be used for anything but its stated purpose. Almost no dispensary can say that about its own customer data, which flows freely between operations, marketing, and analytics. Write purpose limits into your own data governance and enforce them with access control.
4. Tell your patients the difference. Staff will get asked about “the government surveying people about weed.” The accurate answer — a confidential statistical survey is not an enforcement tool, and it is not the same as a social media monitoring contract — is genuinely useful, and delivering it well builds the trust that every other part of your privacy program depends on.
5. Apply the same test to state reporting demands. When a regulator proposes new operator reporting, ask the NHIS questions: what is the purpose limitation, who can access it, is it firewalled from enforcement, and is there notice? Those are reasonable questions and asking them publicly is how the answers improve.
Bottom Line
The CDC wants to ask Americans how often they use cannabis and why, on the country’s flagship health survey, under statutory confidentiality, with sixty days of public comment. That is not the surveillance story of the month. It is the control group.
The surveillance story is what happened on the same day at a different agency, without notice, without consent, and without a purpose-limitation statute.
Operators should support the CDC proposal, comment to strengthen its confidentiality communication, and then use the contrast as the clearest available illustration of a principle this industry keeps having to relearn: the question is never simply whether cannabis data gets collected. It is who holds it, what they may do with it, and whether anyone had a chance to object.
The proposed data collection was published in the Federal Register on August 20, 2026; written comments are due October 19, 2026. Reporting on the 2028 NHIS draft questionnaire from Marijuana Moment and The Marijuana Herald. Confidentiality protections summarized from Section 308(d) of the Public Health Service Act and CIPSEA.
