Research

Reading the field against the rule.

These are rolling, aggregate studies that read publicly posted compliance artifacts against the regulation that governs them. Each study reports the field in aggregate, names no organization, and is reproducible through an open assessment skill. The same roster is re-read on a fixed cadence, so each study measures not just where the field stands but how it moves.

Studies
Study 01 Interim read published
The State of the Part 2 Patient Notice

An aggregate, de-identified read of publicly posted 42 CFR Part 2 patient notices against the 2024 final rule. The findings are published at dangssolutions.com, where the practice that conducts the study is based. The method is described here.

Read the findings →
Study 01 method

How the Part 2 notice study reads a notice.

  • One checklist, frozen. Each posted notice is captured as plain text and read against the required elements of 42 CFR 2.22(b), under a frozen element codebook (C1 to C15, v1.1). Freezing the codebook before collecting at volume keeps early and late reads comparable.
  • Four result tiers. Current (all required elements present), partial (a post-2024 notice with element gaps), outdated (a pre-2024 notice still posted), or missing (no clinical notice posted).
  • A random, stratified sample. The population sample is drawn at random from the enumerable set of SUD treatment programs, stratified so the estimate describes the field rather than the programs easiest to find. Every headline figure carries a confidence interval and its sample size.
  • A measured instrument. A blind validation subsample is coded independently by a human reviewer who does not see the skill's read first. Agreement between the two is what turns a field survey into a measurement: how often the read is right, and how faithful its evidence is.
  • Reproducible. The read runs on the open Part 2 Notice Assessment skill, Apache 2.0. Anyone can re-run it against the same public notice.
How these studies work
  • Aggregate only. Findings describe the field. No organization is named or made identifiable.
  • About the artifact, not the practice. A study reads what a public document says against what the rule requires, never an assertion about how an organization operates.
  • Open and reproducible. The assessment runs on a published, open-source skill. Anyone can re-run it against the same source.
  • Rolling. The same sampling frame is re-read on a fixed cadence, so each release reports the delta as well as the level.
Need it done, not explained

Need this done rather than explained? Dan Gonzalez runs Dang's Solutions, the practice that delivers it.

Read the Study 01 findings