A framework for measuring safety

Any single measurement of safety is likely to be over-simplified. For example, standardised mortality ratios (as a marker of safety) may be unreliable due to their dependence on administrative data and inability to provide a comprehensive overview of safety. There is no single valid measurement of safety and organisations need to use a variety of measurements and methods to evaluate the care they deliver.

Charles Vincent and colleagues have developed a framework with five classes of information to give an idea of how safe an organisation is.


  1. Figure 1. A framework for measuring safety

    1. Past harm: this encompasses both psychological and physical measures.
    2. Reliability: this encompasses measures of behaviour and systems
    3. Sensitivity to operations: the information and capacity to monitor safety on an hourly or daily basis
    4. Anticipation and preparedness: the ability to anticipate, and be prepared for, problems
    5. Integration and learning: the ability to respond to, and improve based on analysis of patient safety incidents

(Figure 1 was reproduced from: The Health Foundation. Practical guide: A framework for measuring and monitoring safety. London: The Health Foundation, 2014. Pages 12–13)

Importantly, this framework does not focus only on past harm, but also current performance, future risk and organisational resilience. It does not recommend any single measure and emphasises that using a single measure of safety is insufficient.

Illingsworth summarises the practical applications of this framework as follows:

  • Managers and frontline health professionals should reflect on the utility of data collected and rationalise measures used or introduce new ones.
  • Board members and senior executives should actively seek information about hazards and risks, and not just react to previous harm caused.
  • Government, regulators and national bodies should encourage organisations to demonstrate safety using both local and centrally mandated measures.