ADDITIONAL FILE 2: DOCUMENTED SEARCH UPDATE
Executed 14 September 2026 (UTC).

The update protocol was saved before the three database queries. This is not an externally registered protocol or a reconstruction of missing historical sessions. The accompanying metadata exports and decisions document the current update.

309 source-query records; 102 duplicate appearances; 207 unique database records screened. Seven empirical reports were retained (two already cited, five added), plus two associated bibliographic records (a letter and a corrigendum). Three additional empirical sources were added from targeted regional/repository discovery. The total addition is eight empirical sources and two associated records.

Screening was a single AI-assisted assessment; independent human duplicate screening was not performed. Evidence access is stated for each retained source. Abstract-only and metadata-only records are explicitly distinguished. The clinical conclusions do not depend on unexamined full-text details.

Supplementary web hits are discovery leads, including duplicate and irrelevant pages, not a complete native SciELO/Redalyc export. Unresolved and incompletely retrieved regional leads are identified in the web log. They are not classified as evidence of absence.

The original 37-reference corpus includes conceptual, methodological, normative, and previously verified empirical sources. Database update counts must not be reported as the selection counts for that historical corpus. Source verification and searches in September replace the conflicting January cutoff statements; earlier July access dates are historical source consultations.

The export CSVs contain bibliographic metadata, not copied full texts. Retrieval logs preserve dates, endpoints, response sizes, and hashes of locally examined responses. Use DOI/PMID links to consult source texts. The Python script reproduces the three database queries; returned counts may change with indexing.
