It's not a trivial claim but neither that hard to prove:
Claim 1: With the question's notation, $\;[N(K):K]=\left|N(K)/K\right|\;$ is coprime with $\;p\;$ .
Proof: Otherwise there's an element $\;xP\;,\;\;x\in N(K)\;$ of order $\;p\;$ (why?), and thus $\;\langle xP\rangle\le N(K)/K\;$ is of order $\;p\;$ . By the correspondence theorem, there exists a subgroup $\;P\le X\le N(H)\;$ s.t. $\;X/K=\langle xP\rangle\;$ .
But then both $\;K\;$ and $\;X/K\;$ are $\;p\,-$ groups, which makes $\;X\;$ a $\;p\,-$ group containing $\;P\;$, contradicting the maximality of $\;P\;$ as $\;p\,$ subgroup of $\;G\;\;\;\;\;\square$ .
Now to our original problem: suppose the order of $\;x\;$ is $\;p\;$ (otherwise simply take $\;x^{p^{a-1}}\;$...) . If $\;x\notin K\;$ , then $\;xP\;$ is an element of order $\;p\;$ in $\;N(K)/K\;$ , contradicting claim 1 above. Q.E.D.