Maybe an example would help.
Let’s say that we validate equipment. In one test event, one of our testers doesn’t follow GDP in a couple of instances. In this case, we would consider these deviations, correct the issues (using appropriate controls), and move on. If, however, that same tester exhibited repeated occurrences of GDP violations or all our testers were violating GDP practices, we would have to consider this a systemic issue. In this case, we would write a CAPA, determine root cause, and correct the issue. Note that in this case, we would have both corrections (correct the GDP violations) AND corrective actions (address the root cause).
I chose testing to show that the concept can be applied across all disciplines. Maybe you assemble 50 widgets and 1 is not built correctly - you’d probably consider this a deviation. If 1 in every lot is not built correctly (with the same issue), you may have a systemic issue and need to assess root cause / corrective actions. Maybe you write software and testing indicates one of the programming teams consistently introduces memory management errors - enough so that it is considered systemic. Probably a CAPA is warranted to understand why and to correct the root cause.
Hope that helps.