Give each result a complete reference frame
A result needs the outcome, groups or condition compared, direction, magnitude, unit, and uncertainty where appropriate. State the analysis population and time point when they are not already clear. Use tables for dense estimates, but make the text identify the findings that answer the research question.
Do not replace an estimate with a significance label. A small p-value does not show practical importance, and a large p-value does not establish no effect. Report intervals and discuss what values remain compatible with the data.
Patterns for precise reporting
Outcome Y was [estimate and unit] higher in [group A] than in [group B] ([interval]), with [relevant adjustment or analysis population].
The estimate was imprecise ([interval]), leaving both [scientifically relevant possibility A] and [possibility B] compatible with the data.
The first pattern requires a real comparison and correct direction. The second helps describe uncertainty without turning an inconclusive result into proof of absence.
Keep design limits visible
Observational results should normally use associated with, differed, or was related to rather than caused, improved, or prevented. See Association vs Causation. Verify every number, unit, table reference, subgroup label, and sign against the final analysis output.