Trusses -as long as hinges at the joints can be surmised- are less affected by displacement of the nodes than rigid joint frames if the displacement are small... basically, the truss has scarcely changed shape and then the axial forces are to be quite similar.
In any case, it is good to always consider P-Delta whatever the class of the structure, and even initial imperfections of various kinds for some structures, such columns welded from 2 segments or top chords of pony truss bridges. P-Delta for the later with initial imperfection out of the truss plane will be very convenient since it will prove the stability of the chord when buckling out of the plane starting from the initial imperfection. So you see there are cases it is convenient to include it, for there are cases where the goodness of the study will be higher doing so.
It has further advantage in steel structures in that a structure calculated with P-Delta and where any modeling segment (beam, column) meets the industry standard of about 1/1500 maximum initial imperfection can be designed directly without any amplifying factor for moments, i.e., directly form the values given in the output of the analysis...you have only to check that no reduction of the bending and axial stiffness is required on account of the standing axial load at the factored level, which will be the case for most members of most structures. There being some members as compressed, simply reduce E in adequate way and you can still proceed the same way. Goddbye to amplifying factors for moments -and separate consideration of member buckling-, this way.
If you use the Eurocodes, even if the initial imperfections of segments in the models (columns, beams) exceed 1/1500 you can proceed equally as long as you use limit stresses consistent with the assumed initial imperfection.
