Comment: probably this idea helps:
Wilson's theorem says:
$(p-1)+1\equiv 0 \bmod p$
Case 1-If (n+1)=q is prime we can write:
$[(n+1)-1]!+1\equiv 0\bmod (n+1)$
We rewrite $n!-(n+1)^k$ as:
$P=n!-(n+1)^K\equiv -1\bmod (n+1)+(n+1)^k\equiv -1 \bmod (n+1)$$\space\space\space\space (1)$
Case 2- If (n-1) is prime we have:
$[(n-1)-1]!+1+1\equiv 1\bmod (n-1)$
$P=1\mod (n-1)-(n+1)^k\equiv [1-(n+1)^k]\bmod (n-1)=m(n-1)-(n+1)^k+1$
considering the fact that:
$n!>(n+1)^{\lfloor \frac{n-1}2 \rfloor}$; if n is odd.
$n!>(n+1)^{\lceil \frac{n-1}2 \rceil}$; if n is even.
Substituting n by (n-1) we get:
$(n-1)!>n^{\lceil \frac{n-2}2 \rceil}$; if n is odd.
$(n-1)!>n^{\lfloor \frac{n-2}2 \rfloor}$; if n is even.
Suppose n is even and $n+1$ must be prime as OP suggest , then problem reduces to:
$$\begin {cases}P=m(n+1)-1; m,intger \\(n-1)!>n^{\lfloor \frac{n-2}2 \rfloor}\end{cases}$$
Update: The application of this theorem to relation (1) may also help:
Lioville theorem:
For prime number $p>5$ and natural number $m$ the equality $(p-1)!+1=p^m$ is not possible.
I saw this theorem and it's proof in this book:
"250 problemes de theorie elementaire des number"
By W. Sierpinsky.