Hi,
With the capacitors 200nF and 1000uF at the 5V line it is almost impossible that those voltage spikes really exist.
Therefore i'd say most if it is an erroneous measurement.
You need a solid GND plane or at least a solid star point with short and wide tracks to the devices.
And you need to setup your scope GND connections to this point.
And after that do a test:
* connect scope GND to circuit GND star point
* connect scope signal to the star point also
* set scope for single shot triggering
* disconnect/reconnect mains
* most probably the scope triggers and shows some signal
This signal is the induced signal into the scope lines. Obviously this is not a true signal.
It may have different sources.(capacitive or inductive influence, but also EARTH loop, maybe scope included)
***
19V on a 5V line may immediately kill some 5V devices.
Because you say the devices fail...there will be problems.
I expect wiring problems (ground loops, long wires, no star conndction.) from your supply to your application.
Also the application PCB may cause the failure.
To better assist you, I'd like to see a photo of the complete circuit (power supply and application and wiring)
Also a photo of the PCB layout can help to find the problem.
Klaus