Loading evidence directory
Loading evidence directory
Exploratory analysis · live database snapshot
The corpus is large enough to reveal patterns in publication, project restatements and evidence gaps. It is not a statistical sample of Australian hospital construction, and most cost figures cannot be compared across projects.
Calculated 29 August 2026 from 809 published single-site projects. Program funding lines, unresolved records and quarantined facts are excluded.
Finding 01 · Coverage
All eight jurisdictions are represented, but Queensland and Victoria hold most of the directly imported official data. In the other six jurisdictions, more than nine in ten visible facts were machine-read from documents.
Share of visible facts, not share of projects.
Machine-read 90.5% · Secondary 9.5%
Machine-read 94.3% · Secondary 5.7%
Machine-read 100%
Official import 80.5% · Machine-read 7.4% · Secondary 12.1%
Machine-read 90.4% · Secondary 9.6%
Machine-read 98.8% · Secondary 1.2%
Official import 71.8% · Machine-read 6.8% · Secondary 21.4%
Machine-read 95.8% · Secondary 4.2%
Distinct projects carrying at least one visible fact in each field.
| Jurisdiction | Projects | Capital cost | Completion date | Construction start | Floor area | Beds | Variance reason |
|---|---|---|---|---|---|---|---|
| ACT | 27 | 2696.3% | 27.4% | 00% | 13.7% | 27.4% | 13.7% |
| NSW | 209 | 14569.4% | 10349.3% | 52.4% | 7033.5% |
Finding 02 · Cost histories
The automated screen found 0 project series in two or more financial-year editions of the same source family. In 0, the first and last stated amounts differ. These are candidates for source review, not a performance statistic.
Bars start at zero. Counts are printed at the end of each bar.
Finding 03 · Metric readiness
A project can carry cost, beds and area while still being unusable for a ratio. The cost may cover one stage, the bed count the whole hospital, and the area a different boundary again.
Useful for finding and tracing a project, not automatically comparable.
A date alone does not establish construction duration.
The minimum fields needed to describe elapsed construction time.
Co-presence only. Bed counts may mean new beds, total beds or points of care.
Co-presence only. Cost scope and area boundary still need to match.
Still not a valid ratio without compatible boundaries and conventions.
Finding 04 · Source reviews
The same-edition audit returned grants, funding components, increments, project totals and expenditure lines to their original tables. Only one reviewed case was a genuine same-metric source discrepancy: an ACT PDF and spreadsheet differed by $10,000.
$100m is the Commonwealth HHF contribution; $310m was the original whole-project budget, later rescoped to $181m.
Read the ANAO Health and Hospitals Fund audit$5.892m of forward works plus $36.3m of next-stage funding reconciles to the announced $42.2m total; the AIP separately reports $41.412m.
Read the East Metropolitan Health Service history$15.6m, $2.275m and $1m are contributions from different program and agency lines that reconcile to an announced $18.9m total.
Read the Queensland 2022-23 Capital StatementAgency-held Next Steps components were merged with a separate $27.778m implementation project. After separation, the PDF and XLSX still differ by $10,000 on one component.
Read the ACT 2026-27 Statement CThat finding is now encoded in the database. Each raw cost assertion has a separate, reviewable capital-quantity interpretation covering whole-project TEI, allocations, contributions, components, expenditure, measurement stage and project boundary. Unreviewed meanings stay visible as evidence but do not enter calculations.
Finding 05 · Place
226 hospitals have at least one analysed project;104 have more than one. Project counts are useful for navigation. Summed dollars are not audited lifetime capital totals because projects can overlap in scope.
| Hospital | Jurisdiction | Projects |
|---|---|---|
| CAIRNS BASE HOSPITALCAIRNS | QLD | 19 |
| ROYAL DARWIN HOSPITALTIWI | NT | 13 |
| ROCKHAMPTON BASE HOSPITALROCKHAMPTON | QLD | 11 |
| CABOOLTURE HOSPITALCABOOLTURE | QLD | 10 |
| LAUNCESTON GENERAL HOSPITALLAUNCESTON | TAS | 10 |
| LOGAN HOSPITALMEADOWBROOK | QLD | 10 |
| ROYAL HOBART HOSPITALHOBART | TAS | 10 |
| MODBURY HOSPITALMODBURY | SA | 8 |
| BARWON HEALTH - GEELONG HOSPITAL CAMPUSGEELONG | VIC | 6 |
| FOOTSCRAY HOSPITALFOOTSCRAY | VIC | 6 |
Finding 06 · Evidence quality
100% of visible facts carry an exact source location, so a reviewer can get back to the page or table. That does not make every extraction correct. Most records are sparse, most project statuses are unknown, and only a minority point to source bytes preserved with a content hash.
2,358 facts
1,773 facts
586 facts
kept for audit, never displayed
Method and limits
Published projects marked as single-site, and facts that have not been quarantined. Coverage counts use distinct projects, not raw fact rows.
224 program or funding-line records, 3 published unresolved records, and 63 facts that failed value checks.
No cross-project cost ranking, cost overrun percentage, cost per bed, cost per square metre, energy score or AusHFG compliance score.
This is a live exploratory view, not a released statistical dataset. Counts will change as records are split, merged, reviewed and reclassified. Read any important result through to the cited project record and original source.
| 3416.3% |
| 188.6% |
| NT | 36 | 36100% | 12.8% | 616.7% | 00% | 25.6% | 00% |
|---|
| QLD | 181 | 17295% | 105.5% | 00% | 116.1% | 84.4% | 10.6% |
|---|
| SA | 69 | 5072.5% | 3550.7% | 00% | 710.1% | 1724.6% | 1014.5% |
|---|
| TAS | 46 | 46100% | 3269.6% | 1634.8% | 24.3% | 919.6% | 715.2% |
|---|
| VIC | 170 | 15289.4% | 5230.6% | 10.6% | 2917.1% | 1911.2% | 10.6% |
|---|
| WA | 71 | 71100% | 1926.8% | 45.6% | 45.6% | 1419.7% | 1521.1% |
|---|
None survived as a like-for-like cost increase.
The $10.222m row funded planning, project definition, enabling works and design inside an already-announced $1.8bn envelope.
Read the WA project pageThe $1m planning line sat beside $168.525m of construction funding. Later reports establish a $169.525m whole-project baseline.
Read the Victorian PAEC responseA small refurbishment/expansion proposal was replaced by a new 32-bed building with a materially broader clinical scope.
Read the Queensland announcementThe $129.995m amount accumulated design and enabling packages before the later $1.471bn whole-delivery commitment.
Read the ACT 2026-27 Statement GA $925m Stage 2 forecast had been attached to the completed $146.3m Stage 1A record.
Read the Capacity Expansion Program reviewThe convention used by AusHFG departmental benchmarks.
No project-level public source has been loaded.
Cost basis
14.4%190 of 1,324 cost assertions state a basis. Without land, equipment, ICT, escalation and enabling-work inclusions, cross-project subtraction is not defined.
Area convention
0145 area assertions exist. 65 Gross Floor Area, 80 Not stated. AusHFG departmental benchmarks use GDA, not whole-project GFA.
Source conflict
0Some are rounding; others are grants, annual expenditure and total project cost merged under one label. They are excluded from the restatement chart and queued for boundary review.
The arithmetic is available sooner than the measurement compatibility.
Source review still rejected examples inside those reduced sets. Whole-project cost was paired with component area; campus area with expansion cost; total beds with net-new beds. The page therefore publishes cases, not a league table.
The four-bed ward is only one output of a $92.8m hospital that also includes emergency, imaging, pathology, dental, rehabilitation and mental-health services.
Official project scopeThe project adds 2,429m² to a 14,288m² existing campus. Dividing the project cost by total-campus area produces an artificially low rate.
NSW assessment reportThe database contains whole-project areas and a 3,300m² single-room scope increment. Pairing the whole-project cost with that component delta is a Cartesian error.
WA Auditor-General report