The real hyperbolic $n$-space is $$\mathbb H^n:=\{x \in \mathbb R^{n+1}: \langle x,x \rangle =-1 \text{ and }x_0>0\}$$
where $x=(x_0,x_1,\ldots,x_n)$ and the bilinear form is $$\langle x,y \rangle = -x_0y_0 + \sum_{k=1}^n x_k y_k.$$
The tangent space at $p\in \mathbb H^n$ is $T_p \mathbb H^n = p^\perp$.
Consider $p \in \mathbb H^n$ and a unit vector $v$ tangent to $\mathbb H^n$ at $p$. Let $\alpha(t)$ be a geodesic starting at $p$ with velocity $v$.
The fact that $\alpha(t)$ is a geodesic means that its acceleration is normal to $\mathbb H^n$, meaning that the curve has no tangential acceleration, in other words, $\alpha(t) \parallel\ddot \alpha(t)$.
There is a natural curve starting at $p$ with velocity $v$ and without tangential acceleration.
$$\alpha(t) := \cosh(t)p +\sinh(t)v.$$
Note that
- $\alpha(0)=p$,
- $\dot \alpha(0)=v$,
- $\ddot \alpha(t)=\alpha(t)$.
Therefore, $\alpha(t) = \cosh(t)p +\sinh(t)v$ is the geodesic starting at $p$ with velocity $v$ (it cannot exist another geodesic with this initial conditions because geodesics are described by a first-order ODE).
Observe $\alpha(\mathbb R) = (\mathbb R p + \mathbb R v)\cap \mathbb H^n$. With that, we conclude that geodesics are obtained as the intersection of two-dimensional subspaces of $\mathbb R^{n+1}$ with $\mathbb H^n$.
In your case, for two distinct points $p$ and $q$, the geodesic containing these points is $(\mathbb R p + \mathbb R q)\cap \mathbb H^n$ and it is unique because two linearly independent vectors $p,q$ define a unique two-dimensional subspaces of $\mathbb R^{n+1}$.