Method

City Health Evidence Finder is a public beta evidence base of city initiatives to improve health.

It is designed to help policymakers, researchers, city leaders and practitioners find what cities have tried, what evidence exists on impact, what powers and policy levers were used, and what lessons may apply elsewhere.

The Finder covers initiatives affecting health, healthcare, quality of life and the wider conditions that shape health.

Records are identified through an AI assisted global scan of public sources, including:

• peer reviewed academic papers
• city and government reports
• evaluation reports
• policy documents
• project websites
• dashboards
• institutional publications
• reputable media sources

The evidence scan is run regularly. The website is updated weekly with new records and improvements to existing records.

Each scan starts by checking existing records for stronger, newer or conflicting evidence before searching for new initiatives.

New records are not published automatically.

A record can be included where:

• there is a named city, city region or municipal authority
• there is a specific policy, programme, service, regulation, infrastructure project, funding approach, governance arrangement or data system
• the city has a clear role
• there is at least one identifiable public source
• the initiative is relevant to health, healthcare, quality of life or the wider conditions that shape health

A record will normally be excluded where:

• it describes only a national policy
• the city role is unclear
• it is a broad strategy or branding exercise without a specific action
• no reliable public source can be identified
• there is not enough information to describe the initiative accurately

New records enter the evidence base as Draft or Review required.

Before a record can appear publicly, it must pass checks covering:

• unique record identifiers and public URL
• duplicate initiatives
• required public fields
• valid source links
• controlled issue categories and municipal levers
• implementation status
• public wording
• dates
• map coordinates or a recorded reason for map exclusion

Records that do not pass remain on Hold. Archived duplicate records are retained in the private evidence base but do not appear publicly.

Only records marked Public beta are shown on the website.

Before launch, the initial database underwent a dedicated improvement cycle covering:

• duplicate and identifier checks
• source reconciliation
• independent taxonomy classification
• independent review of taxonomy decisions
• targeted manual correction
• city coordinate verification
• public text cleanup
• final schema and publication checks

The process created a documented audit trail. It does not mean that every judgement is beyond dispute. City initiatives often cover several policy areas, and the underlying evidence varies considerably between records.

The Finder does not use subjective labels such as “ready to use” or “promising”.

Records instead use factual fields:

• Implementation status: Describes whether an initiative is planned, being piloted, implemented, scaling, completed or discontinued.
• Source type: Describes the source presented to the user, for example a peer reviewed paper, city government report, evaluation report, policy document, project website or media report.
• Impact evidence class: Describes the type of impact evidence identified.
• Source quality tier: Provides a practical indication of the strength of the source. It is not a formal academic risk of bias assessment.
• Verification status: Records the level of source checking completed.
• Needs impact follow up: Shows where further outcome evidence should be sought.

These fields help users judge what evidence is available. They are not formal academic quality scores and do not mean that an initiative has been proven effective.

Records are updated when stronger, newer or conflicting evidence is identified.

Each record has a stable Case ID and public URL. The record page shows when it was last checked and updated.

Where a newer source changes the evidence, the public summary and evidence fields can be revised while the record identifier remains stable.

Keyword search covers selected public fields, including intervention titles, places, health issues, descriptions, municipal levers and evidence summaries.

The map uses one marker per city. Marker size reflects the number of matching initiatives, and the popup shows the exact number of matching records.

Some records do not appear on the map because they cover more than one location, or because a reliable single location could not be confirmed. They remain available through search.

City Health Evidence Finder is designed for evidence discovery and practical comparison.

It is not:

• a formal systematic review of every city initiative
• a complete record of all city activity
• a formal recommendation that an initiative should be adopted
• a substitute for reading the original source
• a formal assessment of causality or cost effectiveness

The evidence is uneven. Some initiatives have peer reviewed evaluations and measured outcomes. Others are supported only by implementation reports, policy documents or project descriptions.

AI is used to help identify, extract, organise and update records. Errors may remain despite the publication checks.

Users can suggest corrections, missing sources, stronger or conflicting evidence, updated information and initiatives that are not yet included.

Suggestions are checked before any change is made. When reporting a problem, users should include the Case ID and a relevant source link.

Finder: Danielli S. City Health Evidence Finder. City Health Evidence. Public beta. Available at: cityhealthevidence.org. Accessed [date].

Individual record: Danielli S. [Intervention title]. City Health Evidence Finder. Case ID: [Case ID]. Last checked: [date]. Accessed [date].

Users should also cite the original source where they rely directly on its findings.