A \emph{coloring} of a graph $G$ is a map $f:V(G)\to \mathbb{Z}^+$ such that $f(v)\ne f(w)$ for all $vw\in E(G)$. A coloring $f$ is an \emph{odd-sum} coloring if $\sum_{w\in N[v]}f(w)$ is odd, for each vertex $v\in V(G)$. The \emph{odd-sum chromatic number} of a graph $G$, denoted $χ_{os}(G)$, is the minimum number of colors used (that is, the minimum size of the range) in an odd-sum coloring of $G$. Caro, Petruševski, and Škrekovski showed, among other results, that $χ_{os}(G)$ is well-defined for every finite graph $G$ and, in fact, $χ_{os}(G)\le 2χ(G)$. Thus, $χ_{os}(G)\le 8$ for every planar graph $G$ (by the 4 Color Theorem), $χ_{os}(G)\le 6$ for every triangle-free planar graph $G$ (by Grötzsch's Theorem), and $χ_{os}(G)\le 4$ for every bipartite graph. Caro et al. asked, for every even $Δ\ge 4$, whether there exists $g_Δ$ such that if $G$ is planar with maximum degree $Δ$ and girth at least $g_Δ$ then $χ_{os}(G)\le 5$. They also asked, for every even $Δ\ge 4$, whether there exists $g_Δ$ such that if $G$ is planar and bipartite with maximum degree $Δ$ and girth at least $g_Δ$ then $χ_{os}(G)\le 3$. We answer both questions negatively. We also refute a conjecture they made, resolve one further problem they posed, and make progress on another.