Table 1 Patient safety measurement approaches

Voluntary reporting

Healthcare workers report events when they occur using standard forms

Inexpensive, engages staff in safety culture, can capture emerging risks

Reporting behavior drives event capture

Improved communication of what to report; consistency with other approaches

Mining administrative data

Hospital claims data including ICD codes are used to generate “Patient Safety Indicators”

Inexpensive, ability to aggregate measures at hospital, jurisdiction, and national levels

Code capture is influenced by several factors. Codes are not mutually exclusive nor completely exhaustive

Code capture will be improved; codes are now mutually exclusive and completely exhaustive

Mining clinical data

Electronic medical record data are used to generate “Patient Safety Indicators”

Inexpensive, clinically relevant, flexible

Depends on availability of data; lack of consistency between manufacturers; not built-in

Alignment of indicators with higher order concepts

Prospective observations

Observers track patients along their illness trajectory to identify prespecified events

Clinically relevant and acceptable, adaptable to many environments, does not require information technology

Expensive

Consistency with other approaches, enabling development of machine learning and other AI based approaches to detection