It's worth emphasizing how this boils down to determining a line from a point and its $y$-intercept.
We are given $\,f(x) := P(x+1) = (x\color{#c00}{-2})(x\color{#c00}{+2})Q(x) + y(x),\ \ \ y(0)=6,\ \deg(y) \le 1$
and also that $\,f(2)=P(3)=18,\,$ and we seek $f(-2)= P(-1) = $ constant term of $P(x-1)$
But $\,f(\color{#c00}{\pm 2}) = y(\pm 2),\,$ so equivalently we're given $\,y(2)=18\,$ and $\,y(0)=6\,$ and we seek $\,y(-2)$
But we know since grade-school how to determine a line $\,y(x) = mx+b\,$ given both a point on the line and the $y$-intercept $\,b=y(0)\ $ [ditto given a point and the line's slope $\,m = y'(0) = y'$].
Remark $ $ In an abstract algebra course we learn how to view the above problem as a special case of CRT = Chinese Remainder Theorem, i.e. how to compute $\ y = f\bmod (x\!-\!2)(x\!+\!2)\,$ from $\,f(2) = f\bmod x\!-\!2\,$ and $\,f(-2) = f\bmod (x\!+\!2),\,$ e.g. see here. Also we learn the close relationship with various interpolation methods (Lagrange / Newton).