A proxy outcome changed: did the study measure quitting itself?

Write measured and represented on separate lines

Cochrane cautions that surrogate measures may not predict clinically important outcomes accurately. The reading task is to preserve the distinction, not decide from a headline whether a particular proxy is validated. A measure can be useful for one question while being insufficient to establish another outcome.

Locate the endpoint’s definition, assessment method, and follow-up. Then write the stronger claim beside it. If a report measured an intermediate score but a promotion says confirmed quitting, the missing bridge matters. Do not assume that the word scientific supplies that bridge or that a significant change validates every use of the measure.

Cochrane: Cochrane Handbook, Chapter 3: Defining inclusion criteria

Try an original two-box example

Imagine a fictional project measuring an invented readiness score. The authors discuss whether it might help explain later behavior, but the report supplies no measured abstinence outcome. Put readiness score in the observed box and later quitting in the separate target box. Leave the arrow between them marked not established by this example.

The invented score is not a validated questionnaire or a recommended threshold. The example does not claim that readiness can never matter. It shows why a measured change cannot be silently renamed as an outcome that was not observed. Counting each improved score as a person who quit would add information the fictional project never collected.

Look for the relationship evidence, not just correlation

Read what the report claims about the endpoint’s purpose and the supporting evidence it cites. A proposed intermediate marker, a process indicator, and a formally validated surrogate are not labels to exchange casually. If you cannot evaluate the relationship, keep the evidence question open and seek qualified methodological interpretation.

If the target outcome was also measured, inspect that result separately with its own definition and uncertainty. Do not assume it must match the proxy’s direction or magnitude. Conversely, an unresolved proxy claim does not erase every useful observation in the report. The limitation belongs to the inference being made from one endpoint to another.

Share the endpoint without upgrading it

Your note can say: this report found a change on this specified intermediate measure, and I have not established the corresponding target-outcome result. That sentence preserves useful information without making a clinical promise. If a summary already conflates the measures, compare it with the actual endpoint before forwarding the claim.

For individual care, discuss the whole evidence with an appropriate professional. Personal goal completion, support attendance, and a craving note should each keep their actual meaning too. They may guide a practical conversation without proving abstinence or medical recovery. The output of this exercise is a clearly labeled measurement and an explicit limit on what it can represent.