Yes, it is correct for small delta(T). But, as dn/dT also depends on T, this dependence will introduce an error in calculated value of n when delta(T) increases. It is the same as use linear approximation instead calculation of integral. To decide, is delta(T) small enough for correct linear approximation you must know the values of dn/dT within temperature range you consider.
It depends on how detailed you want to describe the phase distortion. " n+dn/dT*delta(T) " works fine in the zeroth approximation. You can find more information from these articles:
Article Thermal Effects in a Nd:YAG Laser
Article Thermal Lensing in a Nd:YAG Laser Rod
Article On thermal effects in solid-state lasers: The case of ytterb...