Table 2.

Measuring and understanding hospital mortality

QuantitativeHospital level standardised mortality ratios (HSMR, SHMI etc)
  • Readily available for most hospitals

  • Highly dependent on coding

  • Low rates may give false assurance

  • Lack clinical credibility

  • Do not usually indicate problems in specific clinical areas

Condition or pathway specific standardised mortality rates
  • Readily available for some conditions

  • Clinically credible

  • Robust risk adjustment models

  • Can be viewed alongside other process and outcome measures

  • Not available for all conditions

  • Value limited to specific conditions

  • Not usually integrated with other hospital measures of quality and safety

QualitativeRCRR of all or a sample of deaths
  • Can inform wider learning by highlighting areas to focus quality improvement

  • Clinically credible

  • Time consuming

  • Expensive

  • Some decisions can be highly subjective

  • Unstructured approaches unlikely to lead to useful learning

Individual case-note review
  • Helpful for individual unexpected deaths, investigation of serious incidents etc

  • Can be informed by data on other safety incidents, near misses etc

  • A risk that a whole system approach won't be taken if individual deaths are viewed in isolation

  • HSMR = hospital standardised mortality ratio; RCRR = retrospective case record review; SHMI = summary hospital mortality indicator.