It is a consequence of Sullivan's work

<cite authors="Sullivan, D.">_Sullivan, D._, Combinatorial invariants of analytic spaces, Proc. Liverpool Singularities-Sympos. I, Dept. Pure Math. Univ. Liverpool 1969-1970, Lect. Notes Math. 192, 165-168 (1971). [ZBL0227.32005](https://zbmath.org/?q=an:0227.32005).</cite>

that every compact $k$-dimensional real-analytic subset $V\subset {\mathbb R}^n$ is a mod 2 pseudo-manifold: Every $k-1$-dimensional simplex in a triangulation of $V$ is contained in an even number of $k$-simplices. (This is immediate from the main result of Sullivan's paper about local structure of $V$ as a cone over a base of even Euler characteristic.) Now, take the sum of all $k$-simplices in the given triangulation of $V$. This will be a mod 2 cycle. Since there are no simplces of dimension $k+1$, this cycle is not a boundary. Hence, $H_k(V, {\mathbb Z}_2)\ne 0$. In particular, $V$ cannot be contractible. Note that this argument is pretty much the same as in the smooth case, one you have Sullivan's local result. 

Sullivan's paper is freely available [here](https://www.math.stonybrook.edu/~dennis/publications/PDF/DS-pub-0007.pdf).