A useful improvement measure answers a question the team needs to resolve. It should connect the project's aim to observable information, rather than simply fill an available dashboard. The Institute for Healthcare Improvement recommends using measurement to learn whether changes produce improvement and incorporating the perspectives of people affected by the work.
Selecting measures involves two decisions: what matters and how to count it consistently. The categories described in Structure, Process, and Outcome Measures help organize a measurement set. They do not, by themselves, supply a usable definition.
Write the counting rules
CMS describes measure specifications as detailed instructions covering populations, data elements, sources, timing, and calculation. For a proportion, identify the eligible opportunities in the denominator and the subset meeting the criterion in the numerator. Specify exclusions before reviewing results, rather than removing inconvenient cases afterward.
Create a short definition sheet before collecting baseline data. Include the observation period, unit of analysis, source field or observation method, and person responsible for collection. State what happens when records conflict or required information is missing. If using an established measure, consult its actual specifications and version instead of reconstructing the measure from its name.
For a local measure, try the instructions on a few records together. Ask two collectors to explain how they classified the same ambiguous case. Use disagreement to clarify the instructions before interpreting a difference between units as a performance difference. This is a practical preparation exercise, not a claim that a small trial establishes formal measure reliability.
Check usefulness and burden
IHI recommends collecting information that is useful for learning, including quantitative and qualitative data, and displaying results over time. Existing data and an appropriate sampling approach may make collection more feasible. Ask whether the information will arrive in time to influence the next decision and whether collection is taking effort away from the work being improved.
Also ask what the selected measure leaves invisible. A completion field may describe documentation without establishing that a patient understood the information. Pair the central measure with an appropriate balancing measure and consider whether results should be examined separately for relevant patient groups.
Illustrative selection exercise
Suppose a team wants patients to know whom to contact after discharge. This is a hypothetical example. Counting how often a contact-information box is checked answers a documentation question. Asking patients to identify the contact route answers a different question. The team should select the question that matches its aim and explain the limits of the chosen method.
Before using either result, the team could define which discharges are eligible, when the check occurs, and how unavailable responses are recorded. It could also note staff time required to collect the information. Keeping these rules visible makes later interpretation more defensible than choosing whichever measure produces the most favorable percentage.
If the team wants to compare groups, plan how to assess equity in the measurement.
Sources
- IHI: Establishing Measures (accessed 2026-09-17).
- CMS Measures Management System: Develop Specifications (accessed 2026-09-17).