Who enters reported treatment?

Within a consistently reporting state, how do reported treatment admissions’ substance and referral mixes differ by age, and which patterns survive sensitivity checks?

SAMHSA TEDS-A 2024 public-use data: 64,443 Michigan admission episodes. Ohio was assessed first; unresolved coverage and missingness concerns prompted a documented Michigan fallback. Raw records stay in the source workspace.

Descriptive age-group composition with explicit known/all-record denominators, missingness bounds, restrictions by completeness/prior history/service, fixed-service standardization and hypothetical omitted-coverage stress scenarios.

Existing tests.log records 12 passing tests. reproduction.json records byte-identical repeat outputs. independent_audit.json reports a separate standard-library parser check matching national (1,583,823), Michigan (64,443) and Ohio (16,363) totals. This is an independent code-path check, not an external audit.

Older adult admissions have higher alcohol and individual-referral shares and lower court/criminal-justice referral shares. The 55+ versus 18–24 directions persist through tested restrictions and standardization. The opioid contrast reverses in ambulatory-only data.

Episodes are not unique people, population need or clinic demand. Stable reported counts do not prove complete coverage. Individual referral is broader than voluntary self-referral. Bounds and sign consistency are not confidence intervals or significance tests. Aggregate overlapping outputs are for private review, not a formal privacy guarantee.

Codebook-based definitions, descriptive pipeline, sensitivity and standardization tables, charts, independent parser audit, visual report and planning brief.

Trace one denominator and unknown-value rule; explain the opioid reversal and Michigan selection; calculate a missingness bound and weighted example. Write a personal interpretation and seek a qualified coverage review later.

Ask a prospective planning analyst whether this task exists, what decision follows it and whether the available scope adds value. Stop or rescope if the answer is no or the coverage is inadequate. A later trial could populate one requested meeting worksheet and record which questions it informed; no provider benefit is assumed.

What already exists

The Michigan report, planning brief and aggregate sensitivity tables can support discussion of reported admission composition and uncertainty. Practical benefit remains unestablished.

Functional adaptation still needed

First investigate a specific job: preparing the age-by-service-mix section of a Michigan treatment-planning meeting packet to decide which referral-pathway questions need local investigation. Establish whether this is a real task, whether statewide reported episodes fit it, and whether more local/current data are required. Only if appropriate would a later adaptation produce a concise planning worksheet with relevant age groups, denominator definitions and service-mix cautions. The current project cannot forecast provider demand or allocate capacity.

Adoption and outcome evidence

Adoption: requested use of the worksheet in an actual planning workflow, documented with permission. Outcome: whether it identifies a concrete, appropriate follow-up question or reduces preparation time compared with the prior process, while avoiding misleading coverage claims. Track rejected conclusions and required supplemental data. Do not infer better treatment access, capacity decisions or patient outcomes.

The user supplied or authorized the project goal and scope. AI substantially produced the initial code, tests, analysis execution, charts and documentation. This case-study draft is also AI-written. No completed independent user coding, source audit, beneficiary review, adoption or measured impact is recorded. Passing software checks does not establish subject-matter validity. Update this wording only with evidence of work actually performed.