Measurement for Improvement

Read here for a brief overview of Measurement for Improvement. For a more detailed explanation with worked examples, head over to the Education module.

Most clinicians should be familiar with measurement for research, where most commonly two (or more) different treatments are compared to each other. The primary goal of the research is to demonstrate whether there is a difference in outcome 'before and after' between the different treatments, and would usually result in new knowledge about the treatments being studied. The study would typically involve a large (for statistical purposes), selected (and randomised) group of patients, who follow a tightly controlled pathway of care, so that if a statistically significant difference is seen, that difference can reasonably be attributed to the treatment concerned.

When data is used to assess performance against a set standard or 'target', i.e. as part of a quality assurance process, this is measurement for accountability/judgement. This sort of data might also be used to 'benchmark' performance against other departments or organisations.

Measurement for improvement is concerned with how best to demonstrate that the change being made to a system results in the improvement being sought. It involves a much more dynamic approach to measurement, monitoring outcomes and processes within a system over time, offering the opportunity to adapt approaches, building on those that appear to be resulting in improvement.

Key to a successful improvement project is deciding what you want to measure that will best demonstrate the improvement(s) that you are trying to make.

When deciding what to measure, you should consider including measures from each of the following three different 'types':

  • Outcome measures - demonstrate the end result of the change(s) to a process or system, which in healthcare most often involves how the patient has been affected by the change(s);
  • Process measures - measure how the processes within a system are operating, commonly how well (e.g. % compliance with protocol) you are delivering a change that you want to make.;
  • Balancing measures - look for potentially unwanted or unintended consequences, which might have been introduced elsewhere in the system as a result of the change(s) you are making.