Health Systems and Policy Process · Lesson 4

A Testable Theory of Change

5 min read

Lesson objectives

  • Link policy inputs, activities, outputs and outcomes through testable assumptions and a stopping indicator.
Reading tools

A theory of change explains why an intervention is expected to produce an outcome and where that expectation might fail. Begin with a defined problem, population and desired outcome. Link resources and activities to immediate outputs, changes in service use or behavior, and longer-term results. Allocating an appointment slot is a resource or operational act; offering it is an output; a person attending and receiving care is a nearer result; better health is a later result requiring additional time and evidence. These levels are not interchangeable. Extra slots might miss the delayed group or deliver poor-quality care. A diagram does not prove effectiveness, but it prevents the leap from 'we opened evening hours' to 'population health improved' without a mechanism and a measure. Express each arrow as a testable proposition, not a decorative line.

Place assumptions beside the arrows. If slots move to daytime, people with an evening transport barrier must be able to see and book them; that assumes the message reaches those people and booking rules admit them. If slots are booked, visits must be completed; that assumes staff and supplies remain available and a new bottleneck does not arise downstream. Rank assumptions by their importance and how they could be checked before scaling. Ask service users about barriers with appropriate privacy, inspect operations logs and test alternative explanations. An administrative metric cannot establish a patient-health effect. WHO's evidence-informed theory-of-change guidance starts from the problem, outcomes, intervention and mechanism, then validates and revises the model. The value here is discipline in reasoning, not a universal diagram that can be copied across countries and services.

Measure different links in the chain. An implementation indicator counts slots actually opened on schedule. An access indicator might be the share of eligible requests receiving a suitable offer within seven days. Completion counts services finished among bookings. An equity indicator compares suitable offers between those reporting a transport barrier and those without it. Every measure needs a numerator, denominator, period, source, reporting frequency and owner. Add a displacement or harm measure: moving slots from evening may lengthen waits for people who cannot attend in daytime. Set a review trigger for this decision rather than presenting a local target as a universal standard. If a group worsens, examine the cause and change the mix or pause the move. Do not attribute an outcome change to the pilot before considering simultaneous changes in demand, staffing or recording; causal evaluation asks a different question from implementation monitoring.

Lesson summary

Link resources, activities, offers and completion through explicit assumptions. Monitor harm when one group improves at another’s expense.