Gromov, in his monograph (see page 83 of "Essays in group theory"), writes the following. 

> The idea of hyperbolicity has been lingering in combinatorial group theory since the basic work by Dehn. An extensive study of a class of word hyperbolic groups $\Gamma$ with $\mathrm{dim}\, \partial \Gamma = 1$ (in the combinatorial disguise) was conducted by Olshanski (see [Ol]). Deep algebraic results on general hyperbolic groups are contained in the as yet unpublished work by I. Rips who calls them
groups with negative curvature.

As you note, Gromov's earlier papers contain relevant work.  Gromov also refers to Cannon's work, but only as regards rationality of certain languages.  I believe that Cannon's treatment of quasi-geodesic stability, and its relevance for hyperbolic groups, slightly predates Gromov's.  

Both Cannon and Gromov are alive and both reply to email.  So perhaps it would be better to contact them directly.