Methods

Data are drawn from two public Alberta Health Services (AHS) sources: archived snapshots of active temporary service disruptions and the AHS notice/news archive.

The snapshot archive provides recurring records of disruptions active on specific dates. The notice archive is used to identify additional disruptions that may be missed or underrepresented in the snapshot stream.

Intervals were reconstructed from public wording using rule-based parsing. Explicit start and end times were retained directly from the source text.

Recurring overnight, weekend, or reduced-hours schedules were converted into intervals when the wording supported a defensible interpretation. Records with incomplete or ambiguous timing were retained for review but not converted into closure hours unless supported by another source or a manually documented fixed interval.

Repeated records from snapshot pages were deduplicated into episodes using site, timing, service, and text fields. Notice-derived episodes were reconciled against the snapshot-derived corpus to identify duplicates, overlaps, and distinct events not captured in the snapshot stream.

Overlapping intervals were merged before calculating site-year burden.

Manual fixed intervals were added when a disruption was clearly documented in the public source material but was not captured in usable form by either archive scrape. These additions were logged and retained as auditable fixed intervals in the dataset.

The dataset can be refreshed when new AHS archive snapshots or notices become available. Cached reruns support parser or manual-adjudication updates without re-scraping. Full fresh scrapes are required when newly posted public records must be incorporated.

The data-through date for each release is listed in the version history on the SORCTracks page.

SORCTracks is a public-record reconstruction rather than an operational gold standard. Public wording is inconsistent across notices, site names are not always standardized, and some real disruptions cannot be timed defensibly from available source material.

Open-ended disruptions may accumulate hours as the observed source window expands. A disruption with no posted end date will accrue additional hours in each successive data release until a resolution is captured or the disruption is manually closed. This means site-year totals for the most recent period in any given release should be treated as provisional, and cross-version comparisons should account for the data-through date of each release.

Some disruptions may not have been publicly posted at all, or may have been posted but not archived in a form captured by the reconstruction process. SORCTracks therefore represents a lower bound on total disruption burden rather than a complete inventory.

Disruption reasons are categorized from the verbatim text of source postings and reflect the stated reason as posted. They may not capture the underlying or proximate cause in all cases.

The deposit includes documentation sufficient to inspect the analytic logic and verify selected outputs against the public source material. Full independent reproduction of all outputs would require reimplementation of the described procedures or scoped access to additional pipeline artifacts. Source code is not publicly distributed. Researchers seeking to validate specific records or coordinate on independent re-implementation are invited to contact info@sorc.ca.

Analytic data files are embargoed on a per-layer basis pending manuscript release, at which point they will become publicly accessible through the OSF project. Documentation and audit files are publicly accessible now.