Why don't we know whether care is safe?

Reliable data are essential to ensuring that health care is delivered safely and appropriately. Yet the availability of reliable data on safety remains surprisingly poor, as does our knowledge of what it costs (and should cost) to generate such data. The authors suggest the following as priorities: (1) develop valid and reliable measures of the common causes of preventable deaths; (2) evaluate whether a global measure of safety is valid, feasible, and useful; (3) explore the incremental value of collecting data for each patient safety measure; (4) evaluate if/how patient safety reporting systems can be used to influence outcomes at all levels; (5) explore the value-and the unintended consequences-of creating a list of reportable events; (6) evaluate the infrastructure required to monitor patient safety; and (7) explore the validity and usefulness of measurements of patient safety climate.