Outcomes & measurement

A dashboard can hold many numbers and still leave an organization unsure what to do next. Useful measurement begins with a decision: whether to adjust an invitation, improve a handoff, continue a service, or invest in a larger evaluation.

The CDC Program Evaluation Framework, 2024 (opens in a new tab), published September 26, 2024, provides a practical structure for planning evaluations and using findings. It distinguishes evaluation, which supports decisions about programs, from research aimed primarily at generalizable knowledge. Naming the purpose helps a team choose an appropriate question and method.

For a care engagement program, we recommend distinguishing the following:

• Reach: who was offered the service and who could access it.
• Delivery: whether the planned activity actually happened.
• Experience: how people described its usefulness and burden.
• Outcomes: whether a defined aspect of life, care, or service use changed.
• Resources: the staff time, funding, and other inputs required.

These categories are an editorial organizing aid, not a new validated measurement scale. A specific measure may require specialist selection, established scoring rules, permissions, or a suitable comparison.

A message being opened is a delivery observation. It does not establish understanding. A favorable satisfaction answer is an experience report. It does not establish reduced hospital use. Keeping these distinctions visible lets a small result remain useful without becoming an oversized promise.

Imagine a hypothetical community program introducing a clearer family update. Before launch, staff and families agree that the first question is whether recipients can identify the intended next step.

The program could review a sample of updates and invite recipients to explain what they understood. It could record how long staff spend preparing and answering messages, and note when a family prefers another channel. The team would agree in advance how it plans to use those observations.

The exercise may identify a confusing phrase or a missing contact route. That is valuable operational learning. It is not, by itself, a clinical trial or proof that the new update caused a broader health improvement.

The AHRQ Care Coordination Measures Atlas (opens in a new tab) can help teams consider whose experience a coordination measure captures. The professional sending a referral and the family waiting for help may describe the same process differently.

When reporting a number, state the group and period it describes. Include the denominator, explain missing responses, and avoid treating people who did not answer as if their experience matched those who did. Where comparisons are involved, describe other changes that could explain the result.

A local before-and-after observation can suggest a useful direction. Stronger causal claims require a design able to address alternative explanations. The appropriate design depends on the question, the setting, and the consequences of getting the answer wrong.

The DaSH adult day services study (opens in a new tab) examined caregiver experiences around service days. It did not evaluate a software-generated summary. This distinction matters whenever a technology supports a service with an existing evidence base.

The genusConnect Day Center research collection separates claims about services, family communication, software, and implementation. Our proposed measurement approach carries that separation forward: say what was studied, identify what changed, and reserve broader conclusions for evidence that can support them.

An evaluation should end with something people can understand and use. Share what the team learned, the limits of that finding, and the decision it informed. If the result was mixed or inconclusive, explain what the next review will examine. An honest account of uncertainty can guide a better decision than a single impressive number.