# Patefacta readiness update — 12 September 2026

The frozen legacy-history acquisition is complete and its archive is published.
This does not establish complete coverage of every government record or an
independently validated clinical product–study connection.

| Area | Verified result |
|---|---|
| Legacy narratives | All 1,704 frozen-index records captured and verified in production storage |
| Individual responses | 1,697 identity matches; seven wrong-record responses excluded |
| Alternate portal sources | Six matched captures recover factual history for six of those seven cases |
| SBD milestones | Available for all 619 legacy SBD records across the retained API and portal versions |
| Exact DIN coverage | 1,226 bulk records have usable DINs; the separately verified Maviret portal adds one |
| Epsolay date | NOC 35909 reports 2025-08-27 for DIN 02560720; the RDS date remains blank |
| Date distinctions | Separate NOC, SBD issue and activity-table update notes for Rosiver, Maviret and Mektovi |
| Presentation scope | Two Maviret activities explicitly naming DIN 02522470 remain other-presentation context |
| Review controls | Prepared-file import compatibility repaired; unconfirmed writes lock decisions until reload |
| Search startup | Submission can begin while the background status request is pending; each search still verifies the release |

The original modern portal checkpoint remains 671 reviews. Legacy and modern
versions may refer to the same submission, so these counts must not be added
as a count of unique approvals. RDS00835 retains its correct bulk narrative;
its wrong individual response and unavailable portal do not supply extra facts.

## Measured production responses

Three serial SBD-history checks, before versus after archive activation:

| DIN | Before | After | Coverage change |
|---|---:|---:|---|
| 02467550 — Maviret | 3.76 s | 2.32 s | No exact-DIN result → source-bound review with portal history |
| 02516918 — Retevmo | 2.56 s | 1.15 s | Modern review → modern plus archived source tables |
| 02440342 — Rosiver | 2.59 s | 2.01 s | Unavailable matched record → recovered review and portal tables |

All after-history checks returned HTTP 200 with no unavailable records. These
are small sequential observations with different response coverage, not a
controlled speedup estimate or a capacity test.

After import ended, all 18 search requests succeeded. The 15 reused-connection
requests had median 1.016 s and p95 1.451 s. Three new connections took 5.81–7.25 s,
including connection/proxy overhead. An earlier run during import contained a
20.95 s new-connection outlier; that result is retained in the report. Server
search work rounded to 0.0 ms in the final warm run. These measurements do not
establish actual phone rendering speed or a production latency guarantee.

## Verification and limits

The verified build and type checking passed; all 364 automated tests passed. Results are recorded in
[verification.json](verification.json). Local Chromium checks covered 320, 390
and 768 CSS-pixel layouts. A source-component fixture checked Maviret’s date
labels, retained original DIN field and other-presentation warnings. Live API
checks and local component checks are separate; no real-phone or participant
study is claimed.

Still requiring external evidence or independent work:

- Independent human decisions on 25 prepared packets; no human approvals were
  fabricated or recorded by this update.
- A source-supported 150 mg trial-presentation to 420 mg Canadian-presentation
  bridge for the two Adheroza cases. Regulatory citations are now visible
  separately, and the stale PK/PD instruction for BC01 is clarified.
- A representative real-user/phone study and independently labelled held-out
  accuracy evaluation. AI development checks are not clinical accuracy.
- Reconciliation of changing modern government index totals. Missing dates
  and unlinked DINs remain explicit rather than being guessed.

The [acceptance protocol](acceptance-protocol.md) states how to complete those
checks. Full raw captures and review packets remain private. Source is saved
through Sites; the separate GitHub mirror has not been refreshed because of
the previously recorded automatic approval-review block.

Evidence: [archive receipts](legacy-source-coverage.json),
[portal facts](portal-factual-supplements.json), [date notes](portal-date-notes.json),
[history before](history-before.json), [history after](history-after.json),
[search baseline](search-baseline.json), [search during import](search-after.json),
[search after import](search-after-import.json).